2% mortality from coronavirus? I don't think so.

[The_jackalope]

A lot of us have been arriving at a mortality rate for Coronavirus by dividing deaths by infections. For instance, the current numbers are: 1,112 deaths / 44,789 cases for a "mortality rate" around 2.5%.

I was just looking at the coronavirus tracking page https://www.worldometers.info/coronavirus/coronavirus-cases/ and something clicked. The number we should be looking at is the number of resolved cases that died and the number of resolved cases that recovered. The disease has now been active long enough to allow us to build up a statistically representative sample of resolutions, both recoveries and deaths.

I realized as I looked at the numbers on that page that every case will eventually move from the "active" to the "resolved" category. At that point, ceteris paribus of course, the ratio of deaths to recoveries will be about the same as it is now, but applied to the full population of infections. And that number is not a comforting 2%

In fact, out of a total infected population of 44,789, 5,641 cases have now resolved. Of those, 1,112 have died and 4,529 have recovered. The other 39,148 cases have not yet resolved - they are still sick.

It looks to me as though by the time the illness burns itself out, the death rate will look more like 20% than 2%. That puts it firmly in the category of the 1918 Spanish flu.

It's no wonder the epidemiologists are freaking out.

[Fresh_Start]

1. I saw that too...but the died versus recovered...probably is only the people

admitted to hospitals...
the people who recovered at home because it didn't compound into life threatening situation are probably not in the recovered numbers. I'm praying recovered at home is a big number

[The_jackalope]

2. Yes, we don't know how inclusive the numbers are.

But we have a big enough sample size by now to lower the potential variability quite a bit.

[eallen]

8. Yep. Mortality likely is less than 2%, because many people infected likely never get seen.

[Demsrule86]

20. If that was true, there would be no need for quarantine...keep supplies, money and water in your

home just in case...I am. The swine flu was a nothing burger but entire cities are now quarantined with this thing. And we can't count on China to be truthful.

[Crunchy Frog]

19. It's my understanding that a number of people aren't being

either diagnosed or hospitalized, even if they're extremely ill, because there may simply be no space in the hospitals, and not enough testing kits. There are likely fatalities that aren't being registered.

Apparently many people avoid the quarantine facilities because of poor conditions and little or no care for people who get sick.

We have no way of knowing what the real numbers are for any of this.

[Demsrule86]

21. I hope you are right. The other important thing is this...will the virus be self limiting...as bad

as ebola is and was...it is self limiting.

[dewsgirl]

3. I have been following this story very closely and yes the

mortality rate, when all is said and done will be much higher. Something the media doesn't touch is the extremely low recovery rate.
I'm tired of these flu comparisons, you don't lockdown more than 400,000,000 people, because of the flu. I saw something this morning where in the early days of the Spanish Flu, guess what the newspaper's did? .....compared it to the flu.

[SWBTATTReg]

4. When news of this first came out, I heard up to a 10% mortality rate...a horrible figure...

perhaps no longer 10% as earlier reported, but then again, any mortality is horrible. And what actions has our commander in chief dictated? Restrictions on travelers, put other travelers in isolation for some amount of time, and ... what else? No running around getting more funds to those researchers to spearhead a cure if possible, what else? I haven't heard of any other steps undertaken. The one big one prior to the coronavirus, is rump cutting the CDC's budget. Smart.

[Newest Reality]

5. That's

That's interesting and also disturbing.

So, you are predicting at least a one out of five mortality rate there. That is high.

Thanks for your perspective. I wonder why this is not being made more obvious generally, though. Preventing panic? It seems that it would be more important to underscore the gravity of the situation in order to incite more caution, even to the extreme.

Is there more information or maybe links on this kind of extrapolation at this point that you could share? That looks a little too obvious to not be recognized more generally at this point.

[The_jackalope]

10. Yes, at this point it looks like 1 in 5 to me

And yes, it looks obvious, but nobody wants to believe 20%. What struck me as well is that the ratio of serious/critical cases to mild cases is the same as deaths to recoveries- 20%. I don't know what that means at this point, but it looks ominous.

I don't have any other sites I follow yet.

[Newest Reality]

13. Thanks!

Thanks for sharing your observations and calculations, as ominous as your findings appear to be.

Yes, 20% is an overwhelming number to consider.

[Drahthaardogs]

6. No.

Dear God just no. They are freaking out because the R-naught is 3 to 4, which is ridiculously high. The resolved number lags due to reporting.

The death rate is probably lower. There are likely MANY mild cases not reported.

[Blue_true]

25. The leading expert in the world on Coronavirus agrees with you.

He believes that the Wuhan virus should be compared to a bad common cold rather than SARS or MERS, which he stated were 8-10 times deadlier. The guy is a professor at a university in Hong Kong and was part of the team that solved the SARS virus, it didn't say whether he also worked on MERS.

His belief, if I quote the article properly, is that once all people that were infected is known better, the mortality rate of the Wuhan virus will be considerably lower than 2%.

[Drahthaardogs]

29. I am a toxicologist, so even though I only dabble a bit in

Epidemioligy, I am familiar with it in a professional way.

[Amishman]

7. It's very contagious but also very slow progressing

Up to three weeks from infections to symptoms. Usually those who become critical do not do so until two weeks after symptoms appear.

Those dying now were probably exposed almost a month ago.

[spinbaby]

64. I've been noticing that, too

It did not escape my attention that the Brits quarantined in that ski chalet were infected by someone who traveled from Singapore on January 24th.

From what I’ve seen, the typical course of this disease has about a three-week arc: one week incubation, one week mild symptoms, one week severe symptoms. But it’s hard to tell for sure because the Chinese aren’t being upfront with information or perhaps are too overwhelmed to collect accurate data.

[LonePirate]

9. The first US patient in WA state was released from the hospital on Feb. 4 after admission on Jan. 21

I don't know whether he was on death's door in the ICU or if he only suffered mild symptoms.

Given the lack of information and the public panic about the virus, you'd think more information about the virus and general public safety information (wash your hands!!!) was more prevalent.

Also, I read that a vaccine is still about 18 months away. Unless we vastly improve our abilities to create vaccines and ward off future threats, eventually there will be a virus of some sort to come along and decimate the global human population.

[58Sunliner]

40. He was at deaths door with bilateral pneumonia and was given Remdesivir.

He had a rapid recovery and was released 3 days later. It is the drug that Gilead had hoped would stop Ebola. China is testing it now.

[The_jackalope]

43. How available is Remdesivir?

[58Sunliner]

48. Good question. It is considered an experimental drug.

I looked for the source of it, or an analogue-no luck. If it was already available people would be flying/driving to Mexico. I know I would.

[Tatertot]

11. Statistics

“Recovered” is a lagging indicator.

Someone test positive and dies a day later vs someone test positive and takes 2 weeks to recover

I’ve been watching the ratio of died vs recovered for 2 weeks now. It was 60%, now under 20%. It will continue to fall as more recover.

[The_jackalope]

12. I sure hope so. Nt

[PSPS]

14. And your medical degree is from where?

[The_jackalope]

15. My degree is in computer science.

That just means I can do long division with the help of a calculator.

Please feel free to think for yourself.

[PSPS]

69. All well and good but it takes more than that to understand epidemiology

[The_jackalope]

74. No, I'll leave that to the epidemiologists.

Out here in amateur-land, I noticed an odd factoid and commented on it.

I'm not going to try and derive R-nought or make quarantine recommendations or try to predict the eventual number of deaths. I just remark on what I've noticed.

A while ago I did make a projection of a possible 180,000 cases by Feb 28 based on a quadratic curve fit. I've backed off from that based on my perception that the curve may be entering a sigmoid inflection. If that is true, I have no idea what the saturation point of the curve will be, or how the death-rate curve might respond.

It's like my response to climate change projections. I think it's very useful for people to wonder how bad things could get, on the premise that forewarned is forearmed.

[LanternWaste]

80. No is pretending to be an expert in epidemiology.

OP noticed a data trend, and as a layperson, simply commented on it. So I'm not really sure where or how your irrelevance fits in.

[Demsrule86]

17. The math is the math and It is probably worse as China has been lying IMHO. I am a

trained Biologist/ Biochemist with a minor in Math/Statistics, Computers and Chemistry also English literature but I digress. I have been really worried for a while now. You don't quarantine entire cities for a 2% death rate. Those of us who studied 1917 have been holding our breath with every flu epidemic...is this the 'one'? There is always 'one' at some point. And this looks very bad.

[Drahthaardogs]

32. You do when its transmitted via air droplets and fomites

And it has an R-nought of 4.

It is riduculously contagious, so even with a fatality rate of 2%, you can expect stupid high infection rates.

2% is a lot if infection rates go over 65% (which I suspect it will). The R-nought is absurdly high

[Demsrule86]

16. That is my take also...and we don't know how honest China is being and how many are dying without

being counted.

[Turbineguy]

18. The saying goes... "You can prove anything with statistics"

At the beginning there were only infections and death stats. Then they added "recovered". So, the deaths number was higher than the recovered. Then the recovered number was higher than the death number. I noticed that the recovered number then became double the deaths. Does that mean that it was 50% fatal at that point? Then the death number became 1/3 the recovered and now is 1/5.

According to the other numbers on that site, a large percentage of those who died, had other serious health issues. This is a normal occurrence with flu as well, hence the vaccine recommendations for older, less healthy people.

As long as this runs, there will be people who die. That statistic is immediately available. Infections and Recovered numbers are fuzzier and therefore the numbers lag.

I think what has epidemiologists spooked is that the disease is communicable prior to symptoms. That makes it hard to control.

I think going forward the infection rise rate will decline and the recovered numbers will rise rise rapidly.

But that's my interpretation of the statistics, and I could be completely wrong.

[The_jackalope]

23. In stats, the validity of conclusions are

related to the sample size. I think the sample size is now big enough to permit more accurate estimates. Things will change as time goes on, but this is what the situation looks like now.

[Blue_true]

22. The leading epidemiologist in the world seem to disagree with you.

The guy works at a university in Hong Kong and was part of the team that solved the puzzle with the SARS virus. He thinks that the Wuhan virus should be compared more to a bad cold rather than SARS or MERS, which he pointed out were 8-10 times more deadly. He also believes that once more is reported, the mortality rate of the Wuhan virus will end up considerably lower than 2%.

[The_jackalope]

24. I will be very happy to be proven wrong. nt

[Blue_true]

27. I believe you will be proven wrong. The indications are already showing up.

The daily infection numbers have transitioned from a straight line to a sinusoid. That indicates that changes in behavior among people has started to lower the infection rate, but because people are people, the rate will bounce around as it follows the path of a decaying sinusoid that is pointed downward.

[The_jackalope]

34. I agree that at some point the curve will go from quadratic to sigmoid.

All populations that saturate follow a logistic curve eventually. Only in the early growth curve approximates a quadratic, until the inflection point.
Like you, I see evidence that is happening to the case numbers.
It's not happening yet to the cumulative death numbers, that are still tracking a quadratic with 0.9999 R^2. The R^2 of the case curve is dropping, though.

[Blue_true]

52. If you look at all the data from the questionable Day 1 of the outbreak,

then yes, you would conclude that the growth was quadratic. But Chinese officials really mucked up at the start of this virus, so the true shape of the data near the start is questionable. I chose to ignore the first few days of data due to my concerns about it's validity, that is why I saw straight line growth instead of an exponential like some people were claiming or a quadratic like you are stating. The plot seems to have inflected from a line to a sinusoid, like I pointed out. I agree with you that it is too early to say that the sinusoid will be a decaying one, but that is what I personally believe based upon my projection of what human behavior will be.

One thing that discussing what the data for the coronavirus looks like make me appreciate is that in my career I have seldom worked with data that involved people, it is much easier to evaluate data from a bio-reactor than it is infection data that involves people, so my hat is off to the people that work human population data.

[Azathoth]

72. This isn't random sampling

Early identified cases skew heavily towards the most obvious, severe, and progressed infections. These curves in no way reflect what is actually happening in the population.

[The_jackalope]

73. These aren't such early days any more

First reports were from jan 20 or so. Not that long for sure but enough cases have developed in the last couple of weeks to swamp the early errors you refer to.

[Azathoth]

77. Uh, no. This isn't a theoretical question, it's a practical question

You're talking only a couple weeks *max* of meaningful data that have been frantically collected in a necessarily inconsistent and biased way in a population that is likely on the order of hundreds of millions. Wuhan and the Chinese government are already stretched past capacity and their harsh measures are further biasing the already heavily biased sampling by, among other things, incentivizing people to hide their illnesses until they have no choice but to seek treatment. Moreover, we're talking about an illness that may have a 1-2 week incubation period followed by a 1-3 week symptomatic phase. The current data may not even encompass a single full cycle of infection-disease-outcome.

These numbers aren't close to reflecting what is really happening in the population. We're basically fitting curves to noise right now.

It's going to take many weeks before the data accurately reflect the situation.

[The_jackalope]

79. Based on what you say, I'd expect the curves to look noisy

They don't. They look remarkably clean. The case curve shows changes, but none of the scatter I'd expect from a noisy dataset. The death curve is clean as a whistle.

[cwydro]

39. Yeah, I saw someone on CNN earlier, I think with the CDC, who compared it to the flu.

And not the 1918 flu. I can’t remember his name unfortunately.

[58Sunliner]

50. Well SARS vs this coronovirus.

The doctor who was the whistle-blower of this virus, was hospitalized Jan 12 (symptomatic), and did not test positive till the end of the month, just days before he died. Which strikes me as really odd as I read the patient reports I could find and there has not been any lag, per se, between onset of symptoms and positive testing. I read the initial report on the findings of the virus, the anti-bodies it generates. Unless they were messing with him. Some experts have been theorizing that the infection rate is at least ten times higher. This virus, which is like SARS, it attaches at the same cell site as SARS as well, but SARS took 4+ months to infect @ 322 people. Supposedly animal to human, mutation, human to human. This virus, by all accounts, infects hundreds, thousands, in one month. The SARS virus was initially reported as 305 cases, 5 deaths. Initial death rate @ 1.6%. The death rate, 3 months later, would be revised upward and vary significantly with age. Both viruses can cause bi-lateral pneumonia, and ARDS.

As for SARS transmission-"Scientists in the WHO network of collaborating laboratories report that the SARS virus can survive after drying on plastic surfaces for up to 48 hours; that it can survive in feces for at least 2 days, and in urine for at least 24 hours; and that the virus could survive for 4 days in feces taken from patients suffering from diarrhea (WHO Update 47)."
http://sarsreference.com/sarsref/timeline.htm

I read a report about an apt building that had a broken pipe and they thought the transmission to other tenants of this novel coronovirus was caused by this pipe. Not the flu.

[The_jackalope]

56. Thanks! Nt

[Blue_true]

57. Well. I only stated what the leading expert on coronaviruses stated.

You are welcome to your own conclusions.

[58Sunliner]

75. It's just info. And some pause for thought.

https://www.sciencealert.com/latest-coronavirus-study-suggests-it-can-also-be-spread-through-poop

[Azathoth]

26. Here we go with the unskewed polls nonsense

No, it doesn't make sense to look only at the "resolved" column.

Resolved cases are cases that have either been admitted to the hospital and then discharged, or otherwise followed up until a definite resolution is reached. The vast bulk of infections are people who have either not been diagnosed at all, or have been diagnosed and quarantined but not necessarily followed up on in a timely manner.

It's true that the very low mortality rate outside of Wuhan may grow because it may take time for those infections to progress to death, but that is balanced out by the fact that resolved cases at the epicenter likely reflect an inflated mortality rate because 1) almost certainly the total number of infections there is being undercounted 2) Wuhan's medical infrastructure is being stressed past capacity right now and 3) the Chinese government's draconian quarantine measures in Wuhan inadvertently may be producing worse outcomes for those diagnosed there.

So no, the true death rate is almost certainly not 20%.

[Blue_true]

28. I agree with you completely. nt

[TwilightZone]

63. Thanks.

We're doing no one any favors by posting unsubstantiated claims like this.

[Blues Heron]

30. So going by todays numbers, lets say for the sake of argument

Lets say nobody else gets the virus going forward from today - just the people that have gotten it so far that we know about 40-50- thousand - your saying that 9000 more of them will die from this?

[The_jackalope]

46. Not "will die"

"Might die." I think the final death rate will be somewhere between 2% and 20%. I'm quite prepared to consider values at the upper end of that range.

[Blues Heron]

49. I think your using the wrong number for people that have recovered

your number is people that had it bad enough to be hospitalized - the people for whom it was mild don't show up in your stat.

I doubt there are mass graves of uncounted dead either.

[The_jackalope]

51. I'm just using the published number.

The "real" numbers will all be different, but I have no reason to speculate beyond the published numbers.

[Blues Heron]

55. The published number isn't an estimate of people that have recovered

It's people that have recovered out of treatment. That number is never going to include mild casees right?

[The_jackalope]

58. So, what would a more accurate number be?

The death numbers are people who have died in treatment. They might be significantly under-reported as a face-saving measure. We have no certainty on anything yet.

I'm throwing out my number (and its admittedly thin support) mostly to remind people not to ignore possible worst-case scenarios just because they're bad.

[Blues Heron]

65. it's got to be vastly more

All the people that had it mild, thought it was the flu etc.

I think it's good your bringing this up definitely, thanks for staying on top of it

[Ms. Toad]

70. There's only one datapoint I know for sure -

The Washington patient.

The date the recovery was publised in the recovery statistics was 3-4 days AFTER discharge from the hospital.

That could mean there is a lag time in reporting
That could mean there was some point later when he was determined to have been recovered

If it is the latter - I'm still having difficulty believing that there is a systematic means for tracking recoveries (beyond release from the hospital). So my guess is that recovery = release from hospital (perhaps whatever other recoveries are reported). Of the three numbers reported (confirmed, dead, recovered), I consider recovered the most likely to be inaccurate.

[MontanaMama]

31. I went on the USPS website today

to do some international shipping for my business. There was a banner at the top of the Click and Ship page that says:

ALERT: USPS WILL BE TEMPORARILY SUSPENDING THE GUARANTEE ON PRIORITY MAIL EXPRESS INTERNATIONAL DESTINED FOR CHINA AND HONG KONG, EFFECTIVE MONDAY, FEB. 10, 2020, DUE TO WIDESPREAD AIRLINE CANCELLATIONS AND RESTRICTIONS INTO THIS AREA.

Slightly off topic from your OP, but this was unsettling to me.

[58Sunliner]

33. Another factor is the ability to provide the necessary treatment to those who need it.

But some experts think it is grossly under-reported. That is my hope.

[The_jackalope]

35. Either the infections or the deaths may be under-reported, for different reasons.

Either error will change the prognosis, but in different directions.

[58Sunliner]

41. I think they both are.

[pnwmom]

36. How do we know there aren't many more unidentified cases that are mild and never get diagnosed?

[The_jackalope]

37. See above. nt

[pnwmom]

38. Above where?

[The_jackalope]

42. Above your post. I commented that under-reporting can be either positive or negative

Depending on whether it's infections or deaths that are being under-reported.

[58Sunliner]

45. We don't and we won't. Some experts think the # of infections are much higher.

[The_jackalope]

44. A good article by an epidemiologist in NYT

https://www.nytimes.com/2020/02/10/opinion/coronavirus-china-research.html

Very sensible and nuanced. He makes it clear that we are in the early stages of understanding what's going on.

Gabriel Leung (@gmleunghku), an epidemiologist who studied SARS and managed the response to the swine flu pandemic in Hong Kong, is founding director of the World Health Organization Collaborating Center for Infectious Disease Epidemiology and Control and dean of medicine at the University of Hong Kong. He is an adviser to the Hong Kong and Chinese governments on the new coronavirus.

[58Sunliner]

54. Another site-no paywall.

https://247sports.com/college/auburn/board/104012/Contents/the-urgent-questions-scientists-are-asking-about-coronavirus-143665380/

[The_jackalope]

60. Is he the "Gabtiel" you mentioned earlier? Nt

[58Sunliner]

76. I don't remember Gabtiel. Was it in a link?

[Demonaut]

47. I figured about the same a few days ago, I've also screen shot the maps and rations on this site

and another since the 30th of Jan

[Tatertot]

53. "Total recovered"

There is no “resolved”.
There is
“Total deaths” and “total recovered”

Along with “total confirmed”

I agree that you can’t trust the Chinese data at this point.

[The_jackalope]

61. Death and recovery are both resolutions

[Tatertot]

67. not the same

unlike "resolution", "recovery" doesn't include the dead

[The_jackalope]

68. My point was

"Resolved" consists of "recovered + dead". Of those, 20% of the resolved cases were resolved by death, and 80% were resolved by recovery.

[Tatertot]

71. my point

And my point is that the statitics cited by WHO, etc are "recovered" not "resolved"
Some are not using/citing the data correctly as you are.
Currently 1114 dead, 4699 recovered

https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6

[localroger]

59. The big key is the fairly long incubation period

Deaths we are seeing today are due to infections from a week or two ago, when there were a lot less of those. Death rate is probably more like 10-20% for someone with a confirmed infection. This is a very bad disease because of the long incubation period, during which it is infectious, followed by a nontrivial death rate. And being a viral infection once you have it there's not much you can do about it but drink lots of fluids and hope for the best.

[The_jackalope]

62. Agree.

[roamer65]

66. Given the fairly large amount of unreported cases, I think the death rate will be more like 1 pct.

The 1918-1920 flu pandemic was at least 3 percent and possibly higher. At least 80 million died from it. The United States lost around 675,000.

[empedocles]

78. From a different perspective; I wonder about the impact of the 5 million who

fled Wuhan and where, before they started shutting the gates?