TSOT013: Stuvid, a simulcast

Jul 08, 2020, 05:52 AM

Tyson teams with his co-host for Kimology 411 - Kim Schultz.
Short of missing population studies, we are left to our own devices as to whether that very high mortality rate predicted - 10 to 30 times that of the flu - panned out.
Tyson and Kim visit the broad subject again, and Tyson steers the deep dive on Kansas numbers, to make the excellent case of that an incredibly conservative analysis of total infection still results in mortality rate being very low in the state. Similar logic should apply to almost any state.

[CORRECTION: The broad point and the logic absolutely holds for mortality rate as we presented and then proofed out. However, early on, I misinterpreted numbers I saw for deaths on a Kansas visual chart on the topic - the bars on the reporting charts did not represent full reporting weeks, they represented days. So when I said most weeks had only one death and no more than four, that was incorrect. This did not affect my interpretation or analysis of overall mortality however, and it is still true that the biggest spikes were not recently]

Tyson teams with his co-host for Kimology 411 - Kim Schultz.
Short of missing population studies, we are left to our own devices as to whether that very high mortality rate predicted - 10 to 30 times that of the flu - panned out.
Tyson and Kim visit the broad subject again, and Tyson steers the deep dive on Kansas numbers, to make the excellent case of that an incredibly conservative analysis of total infection still results in mortality rate being very low in the state.  Similar logic should apply to almost any state. 

We added a lot of impromptu content, but the rough logic presented is below and it was presented in this order. 

Conservatively figured mortality rate:
·         Mortality rate = Total deaths DIVIDED BY Total infected.
·         277 dead in Kansas from this.
·         Very conservative estimate of infection in Kansas, deliberately halving conservative numbers otherwise arrived at: 174,000 (with easy arguments that it is much, much higher than that)
·      277 / 174,000 = 0.001592 = 0.1592%

o   An easy argument (see below) is made that at least 246,500 Kansans have been infected, which brings this mortality to 0.001124, or 0.1124%. And even this may be a very high mortality estimate (see proofs below).
·         These are in the range of the seasonal flu, which is usually pegged at 0.1% or so.
o   Considering I halved the denominator I’d have otherwise come to, mortality rates of well below 0.001, or 0.1%, are just as, or more, probable.
================================================
Proofs below
·         PROOF: Numerator of mortality rate. 277: This is the most known and solid number. This includes deaths not only confirmed cases but unconfirmed but “probable” (though, I ask, at an average age at death of EIGHTY, (at least where it was just a few weeks ago) how is it “probable” when it otherwise mimics other respiratory illnesses that could kill someone near the end of their life. For example, do not eighty-year-olds near end of life die of the flu?)
·         PROOF: Denominator. 174,000 to 246,500 (with arguments for much larger): Because the state has not done a population study, we are left to our own devices to understand the spread in Kansas. That said, we have good info to get an approximate range, that, by any measure, shows mortality to not be at the initial, model-based, alarming estimate. 


  • These percentages of 12 to 17% infection of the population are a good estimate of the spread of this disease now. Here's why:
    • The length of time the virus has been here is in the general range of that six-month flu estimate
    • These time ranges are already in the range of the CDC estimate for the flu, but R0 (spread factor), being much higher for the new disease than the, flu further justifies using the flu infection estimates. 
      • The R0 (spread factor) is much higher for the new disease than the flu.  Estimated from 2 to 2.5, to up to 5.7
      • This difference in R0 means an exponential difference between how quickly it spreads through the population versus the flu.  Examples:
        • Over only two “spread generations,” using the minimum differences between the two (2.0 versus 1.28), it is estimated to spread to 250% as many people as the flu.
          • ((2*2)/(1.28*1.28) = 2.44). 
        • Over five "spread generations" using the minimum difference in R0, it is estimated to spread to 9.3 TIMES as many people as the flu.  
        • At the maximum difference in R0, (5.7 versus 1.28), this is expected to spread over just two generations, to 19.8 times as many people as the flu ((5.7*5.7)/(1.28*1.28)). 
        • Over five generations, One-thousand, seven-hundred and fifty-one times as many people as the flu.    
      • We’ve had many more than five “spread generations” since this new disease made it to the U.S. and to Kansas.
        • Presumably, “spread generation” should be related to average time to symptoms, which is about five days (not the 14 quoted early – even then that was not correct. 14 was the long range, five was average)
        • Even if one doubles that to ten days, rather than five, as a “spread generation” then TWELVE generations have passed since the first confirmed case in Kansas (3/7/20 to 7/5/20 equals 120 days)
          • Twelve generations means infection spread 212 times as much as flu, to as high as 60,800,000 times as much as the flu (of course other factors would have taken over to slow it down by then, so that latter number is only academically high.  
      • Thus, the low end of the above ratios for five generations is a very, very conservative estimate of how much more this is to have spread versus the flu, to help us validate whether flu spread over six months can be used as a conservative comparison.
      • And that lowest ratio – 9.8 times as many people infected as the flu after a mere five generations, more than makes up for the fact that the time the new disease has had a chance to spread could be considered only 2/3 to 11/12 as long, and that R0 might be lower due to social distancing.
        • (Especially considering we’ve had many more than five generations, and the high end of the estimate of R0 is so much higher. Considering the high estimate in either generations or R0, or even a mid-point in either, results in an astronomical difference between alleged spread between this disease versus flu.   
    • To be even more conservative, let’s take half the percentages of the flu spread over six months to figure the denominator for this new disease.
      • Rather than 12 and 17%, we'll use six and 8.5 percent.
      • Applied to Kansas, that means 174,000 to 246,500 infected, out of 2.9 million. 
  • This results in the mortality rate given at the top of the email.  ~0.11 percent to 0.15 percent.

  • Besides making sense independently, this spread to only 12 to 17% of the population, let alone only six to 8.5%, is well within the results of population studies done elsewhere at least six weeks ago.
    • 14% of New York state, at least six weeks ago, and probably eight or more.
    • 21% of New York City.
    • 4.1% of Los Angeles County, reported 4/20, 2 ½ months ago. 28 to 55 times higher than confirmed cases.
    • 6% of Dade County, reported 4/24.
    • 2.49 to 4.16 of Santa Clara County, reported 4/17. Up to 85 times higher than confirmed cases.
  • “But antibody testing is not as accurate as PCR, so we shouldn't do population studies to understand total infections."
    • We already know a good number are asymptomatic. Estimated easily at 50%.  So picking only confirmed cases at the VERY least misses at least half the cases. That is therefore only 50% accurate. Is antibody testing only 50% accurate? If not, it more accurately helps determine infection rate than does confirmed cases.
      • As an aside, it is a myth that anyone infected, even the asymptomatic, are practically just as likely to spread on the virus – look up this phenomenon and viral load. 
        • The virus does not come in in mass quantities – there is little to expel at first. It incubates IN the body. 
        • So those who end up not even having symptoms directly correlate to not having a large viral load build up in their system. 
        • Further, those without symptoms or before symptoms do not yet experience the main way the virus would slough off into the environment – coughing.
    • And we know that a further decent percentage have mild symptoms.  Some decent percentage, undergoing mild symptoms, are going to wait it out and just stay home or stay away from people, and will not be confirmed.
    • Between asymptomatic and mild symptoms and other reasons people use to not get confirmed, it is easy to estimate that something like 75% or more of those infected don’t get it confirmed. Thus confirmed cases probably at most are catching only 25% of cases, which would make the mortality rate appear to be FOUR TIMES what it actually is
    • Is antibody testing much more likely to false indicate than to correctly indicate? If not, then it MUCH more accurately helps determine infection total than does confirmed cases. (and, anyway, if you believe the 16,000 confirmed cases in Kansas are the only ones in 4 months, I have a bridge in Brooklyn to sell you).   
    • PCR testing (nasopharangeal), to reveal active infection, was developed quickly. Though maybe perfectly developed PCR testing is close to 100% accurate, FDA booklets Tyson has posted screen shots of say that they were not able to get a perfect genetic profile, so had to guess a little. As a result, at least for some of the PCR testing, the document admits that:
      • Other members of the same virus family (most of which are very mild comparatively) can result in a false positive.
      • Bacterial infections can result in a false positive.
      • Others.
      • (Unsure if this was improved since first seen around April, but I expect any marketable test that was used in March is still being used.)    

=================================================
Thanks for listening! Please feel free to email me at tysonharley1776@duck.com with requests for appearances, thoughts, feedback, questions, observations, or show ideas! Please like my episodes and follow my show if that is what feels right to you!