Mortality rate calculation

Login or register to post comments Last Post 0 reads   15 posts
Viewing 10 posts - 1 through 10 (of 15 total)
  • Tue, Feb 04, 2020 - 06:24am

    #1
    Brendon_Watt

    Brendon_Watt

    Status Member (Offline)

    Joined: Feb 04 2020

    Posts: 13

    count placeholder0

    Mortality rate calculation

So the mortality rate is calculated as

deaths/cases

but this doesn’t factor in a current epidemic as the time lag for recovery isn’t factored in.

So maybe a better calculation is

deaths/(deaths+recovered) = mortality rate

it would work out the same for a finished pandemic like sars.

But maybe it would allow the time lag factor to be closer as people have died vs people recovered is the same time period. Maybe not perfect as maybe recovery is longer but better than the current 2% bullshit ?

  • Tue, Feb 04, 2020 - 09:52am

    #3
    RebelYell

    RebelYell

    Status Member (Offline)

    Joined: Dec 21 2019

    Posts: 105

    count placeholder1

    Reply To: Mortality rate calculation

Weird – my post seems to have disappeared.  Attmetping to repost:

See my comment here:  https://www.peakprosperity.com/coronavirus-how-bad-will-it-get/#comment-303493

“My best estimate of the case fatality rate is based on four data points:

– the Lancet study’s estimate of 76,000 actual cases in Wuhan on Jan 25th, and doubling period of 6.4 days
– median time to death of 14 days from hospital admission (lost source link), and assumption that the mean is similar to the median
– my assumption of mean time of admission to hospital from onset of symptoms of 5 days
– the suggestion in the humanevents article linked above that Chinese doctors in Hubei privately agree that official statistics are under-reporting deaths by a factor of three

These assumptions lead to the following calculation:

Number of deaths on Feb 2: 1080 (360 x 3)
Number of cases on Jan 13: 19,000 (76,000 / (12/6.4))
CFR: 11.4% (1080 * 2 / 19,000)

Note that I have simply doubled the number of deaths to account for the fact that only half of the eventual deaths will have occurred by 19 days after infection. This math is overly simplistic since the numbers include all cases, many of which had an infection date before Jan 13th. Therefore it’s reasonable to adjust that number downwards to, say, 9% to reflect this.

Note that, if we accept the assumptions above, which are obviously highly unreliable, the CFR is unlikely to be below 5.7%. It’s also worth noting that if the average time to death is shorter than 19 days from initial infection these numbers would improve significantly for each day shorter. For example a mean mortality period of 12 days from infection would yield a CFR approximately half the value I have calculated.”  The converse is also true and a longer mortality period would imply a higher CFR.

Using deaths/recovered is likely also misleading for a number of reasons.

–  most signficantly the “recovered” number is probably the single most likely to be incorrect statistic as it will almost certainly not include all those people who simply stayed home and had mild symptoms and therefore never appeared in the statistics at all

–  it’s also true that the mean period to recovery is almost certainly different than the mean peiod to death which would tend to distort the ratio too.

Another possible calculation might use the recent number of 25% of cases exhibiting severe or very severe symptoms.  But it is not clear whether this is a case raio or a current ratio, and in any case the number is also likely too high as it will again exclude the folks who just stay home and exhibit mild symptoms.  My guess, based on the lancet study and currently reported statistics is that something like 10%-15% of infected individuals develop severe symptoms and maybe half of those are currently dying.

I would expect the rate to be lower in a healthcare system which is not overwhelmed.

  • Wed, Feb 05, 2020 - 09:04pm

    #3

    gyrogearloose

    Status Gold Member (Offline)

    Joined: Sep 08 2008

    Posts: 376

    count placeholder0

    Death rate 10%

I worked with a similar methadology , but used interpolated data off a graph confirmed infections and deaths someone on zero hedge was making and updating.

Initially assumed 7 days from diagnosis to death of 7 days, and got a death rate of 10%

A couple of days later another article listed showed a median time to death from onset of symptoms of 14 days,  but for 70+yrs of 11 days  and 70- years of 20 days/

This data would push my number (with a feel/fudge for the non linearity ) to 15%

Mainstream papers here ( New Zealand ) all state 2.1% and given that they show the total confirmed and death numbers they are running with deaths divided by total confirmed which is VERY  clearly  a wrong methodology.

Regards Hamish

  • Thu, Feb 06, 2020 - 08:48am   (Reply to #3)

    #4
    RebelYell

    RebelYell

    Status Member (Offline)

    Joined: Dec 21 2019

    Posts: 105

    count placeholder1

    Doc just died – implications for CFR

The doc who tried to warn everyone early has just been reported dead.  He appeared to me to be in his thirties.  He first experienced symptoms on Jan 7th, which implies likely infected around Jan 2nd, and he died on Feb 5th.

That’s 34 days from infection to death.  I hope to God that’s an outlier.  If it’s the mean, given a doubling period of 6.4 days we’re looking at a fatality rate of over 30%.

That said, recent reports from the Chinese government have indicated that 80% of deaths to date occured amongst those over 60, and 75% of people over 60 who have died already have a pre-existing condition.  That data would fit with the idea that older, more vulnerable patients die more quickly – which would have positive implications for the fatality rate.

It is also true that this particular doctor is probably a biased datapoint.  My understanding is that eight medical staff were involved in initially trying to get the word out.  I suspect this doctor was singled out for subsequent attention precisely because he became ill – so it may be dangerous to infer too much from his specific case.

  • Thu, Feb 06, 2020 - 09:02am   (Reply to #3)

    #5
    VeganDB12

    VeganDB12

    Status Silver Member (Offline)

    Joined: Jul 18 2008

    Posts: 288

    count placeholder0

    re: Doc just died – implications for CFR

Do you have a link/source regarding the doctor who died?

thank you in advance

  • Thu, Feb 06, 2020 - 09:42am   (Reply to #3)

    #6
    RebelYell

    RebelYell

    Status Member (Offline)

    Joined: Dec 21 2019

    Posts: 105

    count placeholder0

    re: re: Doc just died – implications for CFR

Here is link:  https://www.cnn.com/2020/02/06/asia/li-wenliang-coronavirus-whistleblower-doctor-dies-intl/index.html

 

  • Thu, Feb 06, 2020 - 09:49am   (Reply to #3)

    #7
    RebelYell

    RebelYell

    Status Member (Offline)

    Joined: Dec 21 2019

    Posts: 105

    count placeholder0

    re: re: re: Doc just died – implications for CFR

And note some errors in what I posted above.  The doc was hospitalized on Jan 12th so perhaps he became infected as late as Jan 7th rather than the Jan 2nd I postulated above.  This would be a time to death of 29 days not 34.  If this were the average that would imply fatality rate >20% (but perhaps not >30%) – but all the caveats about the dangers of inferring too much from this data point still apply.

 

  • Thu, Feb 06, 2020 - 10:19am   (Reply to #3)

    #8
    davefairtex

    davefairtex

    Status Member (Offline)

    Joined: Sep 03 2008

    Posts: 3126

    count placeholder0

    re: Doc just died – implications for CFR

This would be a time to death of 29 days not 34. If this were the average that would imply fatality rate >20% (but perhaps not >30%) – but all the caveats about the dangers of inferring too much from this data point still apply.

Looking at this from a strictly political viewpoint, a dead hero about which you can write whatever narrative you choose is a lot more useful to the Central Committee than a live one that runs around whining about how things should change in very specific ways.

IOW the central authority has a lot of reasons to have him whacked.  Live heroes who have specific – and powerful – criticisms about the power structure are a threat to that power structure.

Seriously.  The guy has already proven himself to be “difficult.”  You really want him running around afterwards making all sorts of noise?

  • Thu, Feb 06, 2020 - 10:31am   (Reply to #3)

    #9
    schmidtma01

    schmidtma01

    Status Member (Offline)

    Joined: May 29 2011

    Posts: 22

    count placeholder1

    re: re: Doc just died – implications for CFR

While I would normally agree with your statement, davefairtx, I think this doc’s popularity among the Chinese people makes it more likely that he’ll be considered a martyr. He gives a (young, previously healthy) face to the corruption of the government. I think that may be a bigger threat to the structure (or, if you’re right, it may be a bigger unintended consequence of that decision).

BTW, as a tangential comment on the times we’re living in: two of the first three people I told this morning that this doc had died asked me the exact same question: “Do you think he was Epsteined?”

  • Thu, Feb 06, 2020 - 10:52am

    #10
    schmidtma01

    schmidtma01

    Status Member (Offline)

    Joined: May 29 2011

    Posts: 22

    count placeholder0

    About those unreported cases…

The Great Hope among everyone that the current 2.1% CNN-reported case fatality will prove to be too high rests on the following assumptions:

  1. The official statistics are actually a thing that is real; and
  2. There a thousands upon thousands of asymptomatic or mild cases that have gone unreported.

This article was published yesterday

Previous preprint suggested cumulative cases of 1723 (95% CI: 427−4471) as of 12 January 2020, and 4000 (95% CI: 1000−9700) as of 18 January 2020 based on the aggregated international export cases [5]. Our analysis yielded cumulative cases of 280 (95% CI: 128−613) as of 12 January 2020, and 609 (95% CI: 278−1333) as of 18 January 2020 based on the exponential growing mechanistic in the early outbreak. Although our estimate case number appeared to have a lower mean than those estimated by Imai et al. [5], they are not statistically different.

(Source)

Those authors also calculated an Ro of 2.6.

The fewer unreported cases there are, the less the CFR will be able to drop on a mathematical basis. Another data point to keep an eye on.

Viewing 10 posts - 1 through 10 (of 15 total)

Login or Register to post comments