Ukraine Mystery Flu: ZeroHedge Estimation of the Case Fatality Rate!

• This is an impressive work by the guys at ZeroHedge. The analysis is pretty long but well worth the read. My own personal opinion is that the deaths have been under reported by an order of magnitude ie it should be in the region of 3000 instead of about 300 reported.
   
  DATA FROM THE UKRAINIAN GOVERNMENT
  The Ukrainian government has been releasing daily data updates on the numbers of “acute respiratory infections” , the number of related hospitalizations , and the number of deaths since Oct 28, 2009.  These reported numbers have increased substantially over the past two weeks, from near zero before Halloween to over a million as of today. It is important to remember that these statistics are not laboratory confirmed H1N1 cases (due to the state of the Ukrainian medical system), but we can cautiously assume at this point the vast majority of ARI (acute respiratory infection) in Ukraine with rapid progression to viral pneumonia are caused by influenza A virus. This is because a large subset of these samples are coming up positive of Influenza A, and a smaller subset tested by rt-rtPCR are coming up positive for influenza A/H1N1, or what is more commonly known as ‘swine flu’. This suggests we are dealing with a viral epidemic of influenza.
   
  
     
  This is not ‘swine flu’ as previously known. Statistically, the only way for this to be ‘swine flu’ is if it has mutated (or if the data is wrong). . . We will get to that in a minute.
   
  As I mentioned , this data comes directly from the Ukrainian government. In the past two weeks, ARI have gone from near zero before Halloween to affecting 1.3 million Ukrainians. Over a million people are currently sick with probable H1N1. More importantly, there are over 75,000 total hospitalized at the time of the writing of this article, Monday 11/16. Based on these two figures (75,000 / 1.3 million), approximately 5.6% of those infected with this new ARI end up at the hospital. This is the critical statistic which allows us to estimate how f*cked we actually might be.
   
  
    
  Officially, there are 299 deaths, but a realistic estimate shows this is easily one order of magnitude too low. Realistically deaths are between 2,000 – 8,000 people from this probable new strain of the flu — so far. The number of dead is sure to rise dramatically, as we are only two weeks into this new epidemic. Which brings us to our estimation of the case fatality rate (CFR).   Again, I would love to be wrong about this.  Let’s hope I am.
   
  ZH ESTIMATION OF THE CASE FATALITY RATE
  The current American strain ‘swine flu’ has a case fatality rate (CFR) of 0.1% or less. Seasonal flu has a CFR of 0.1%. The 1957 pandemic had a case fatality rate of 0.5%. The deadly 1918 pandemic had a CFR of 2 – 5%. Based on the Ukrainian government statistics, the new mutated Ukrainian flu has a CFR of 0.61% or higher — that is, 6.1x as lethal as seasonal flu , perhaps more.  Perhaps less if our assumptions are wrong.
   
  We arrive at the 0.61% CFR statistic based on the number of hospitalizations in Ukraine vs. the number of infections. This estimate is only as good as the underlying data, but I have confidence it is in the ball-park (order of magnitude). I consider the official reported number of deaths (299) to be unreliable for political reasons. No one wants to be the country with the lethal flu. Doctors, especially those in Belarus (which is a dictatorship) feel strong administrative pressure to label influenza viral pneumonia deaths as something else, for example cardiac failure. Thus , ARI hospitalizations are the more reliable indicator for what is happening on the ground.  Deaths can be implied from hospitalization statistics, assuming we are dealing with a form of H1N1 influenza.
   
  Last week, the medical journal the Lancet published a meta-analysis of the (old) swine flu confirming a JAMA report that approximately 11% of hospitalized patients who had contracted the (old) swine flu died.  We will use this 11% figure as our “middle case”.  The official government reported deaths (299) are considered “best case”. For a “worst case” estimate, we can assume that 25% of hospitalized flu patients will eventually die. We can use these 11% and 25% statistics to estimate the case fatality rate of the Ukrainian flu. The method is as follows:
   
  
   
  To calculate the “middle case” and infer fatalities, we take 0.11 (11%, the figure from the JAMA) and multiply it by the number of Ukrainian hospitalizations on a particular date, to yield the expected number of ultimate fatalities. This method assumes the vast majority of reported ARI are from H1N1 swine flu. Using the November 15th data, this gives us 8345 expected fatalities. We take the number of expected fatalities and divide this into the total number who are estimated to be infected on the same particular data point (Nov 15). So 8345 deaths / 1.36 million infections , which yields the case fatality rate, which in this case is 0.61%. Obviously this is a rough estimate, but it is probably within an order of magnitude of the real CFR.
   
  Assumptions for this model:
  1) The vast majority of reported Ukrainian ARI are some form of H1N1 flu.
  2) The data from the Ukrainian government for infections and hospitalizations is reliable and is not ‘massaged’ for political purposes.
   
  Implications of this model:
  1) If A1 and A2 are true, then statistically speaking we must be dealing with mutated H1N1 influenza (mutated ‘swine flu’).
   
  In any case, this estimated CFR for Ukrainian flu is over six times higher than seasonal flu. Not good.  Let’s hope this is wrong. Our “worst case” 25% estimate gives a CFR of 1.39%, enough to eventually shut down commerce and society across the planet — but fortunately it is likely this “worst case” estimate is far off the mark. For our estimates here, including the CFR estimate of 0.61% using the data from the JAMA — supposing this estimate was purely based on Ukrainian government statistics, we could consider such an arbitrarily high Case Fatality Rate unreliable , and the product of bad data.  However, in the context of multiple reports from individuals as well as state health authorities of MANY cases of viral pneumonia and thousands of patients in intensive care in Belarus and Ukraine, we consider this CFR estimate to make much more sense. That is, it fits the data on the ground of a worsening flu pandemic with a tendency towards viral pneumonia.
   
  The old swine flu did not cause viral pneumonia, generally speaking. The new apparently mutated swine flu does, to an extent orders of magnitude higher. Thus , there must have been changes to the receptor binding domain , among other genes (perhaps PA, PB2, and NS1).

end