Covid 10 Fatalities - 4/14 Update

I have continued to update my data each day from the Worldometers.info site.  This has allowed me to continue to track the daily growth rate as we approach the peak of both the Total Cases / day and Deaths / Day.  As would be predicted on from a normal distribution, the daily growth rates would begin to decline as you approach the peak, which has happened in both cases.  Finally, once the daily numbers begin to decline, you should be able to take several days prior to the peak and combine them with several days after the peak, to calculated the daily growth rate that should be zero.  Said another way, this grouping should predict a horizontal growth line.  The day in the middle of this range should be the day of the peak.

As you see above, both graphs display horizontal predicted growth curves.  This suggests the curves have both peaked and begun to decline.  I conclude from this that Cases per Day peaked on 4/9/20 and Deaths per Day peaked the following day, 4/10/20.

In my previous post, I projected the peak for Cases would be 4/11/20 and also assumed the Deaths would peak 4 days later.  It now appears from the graphs above that the peak was 2 days earlier than I had predicted and Death peak was 5 days earlier than I predicted.

So my new forecast, which be significantly lower with the peaks coming sooner, is 39,400 deaths.  This equates to a mortality rate of 4.0% as compared to the Global current rate of 6.3%.   (Remember that this number assumes all infected people have been tested which is clearly not the case). This also predicts Deaths per 10,000 people in the US at 1.19 which compares to other countries predicted rates of  0.04 to 7.0.

So, my last assumption predicting this 39,400 total deaths, is just how skewed the final distribution of deaths will be.  Right now, my prediction is based on the right side of the distribution being 10% larger than the left side which is comprised now of all actual results.  It also assumes that there will not be a significant second wave of infections after the economy begins to open up.  I will continue to record actual results to look for the decline rate to again approach zero signaling the end of the pandemic in the US.

COVID-19 Fatalities - How Many ??

After the CDC announced that the Corona death totals for the United States would range between 100,000 and 200,000, I had a call from a friend who questioned me about these numbers.  Knowing my interest and history of assessing numerical results, he wanted my opinion of these death totals.  So as not give my "opinion" of the CDC Forecast, I began my search for tables of numbers for my assessment.  I was able to find many sources of any given day's totals of cases and deaths, but not historical tables.  For this analysis I did find and use graphs and tables (some produced from graphs) at the following sites:  https://www.worldometers.info/coronavirus/ and https://www.arcgis.com/apps/opsdashboard/index.html#/bda7594740fd40299423467b48e9ecf6.

Understand, please, that I am not an epidemiologist nor do I have much confidence in the current testing of an "infected person" as I am sure there are many infected people that have not been tested which would make infection rates between countries quite variable.  However, it is quite clear when someone suspected of dying from Corona, they are dead.  Therefore, the bulk of my analysis was done with reported deaths, as this was the original question asked of me.  I am not using any medical issues in this approach but only the Total Deaths by Day and their growth rates.  I have used the Case Rates by Day to help assess the shape and duration of the pandemic in various countries.  As part of this, I have also used China and South Korea as countries that have nearly completed their "curves".

•  I graphed the daily death rate and cumulative death rate for the Globe and the US.  As these numbers are growing exponentially, I used growth control charts to help assess when the growth rates were stable and when they changed.  The data ranged from late February to April 4, 2020.  I found that the growth rates slowed for both the Global and US on March 25 and have remained consistent to the present time.  The daily compound rate is 9.8% for the Globe and 20.8% for the US.  My graphs follow:  

•  Next I had to determine when the US would reach the peak or apex on daily deaths (about half way through the crisis).  I started by trying to asses the graphs of daily CASES, not deaths, by country  from the John's Hopkins Dashboard.  Here are few country examples:

China with a double peak but "finished"

South Korea, with extended skew and nearly "finished"

Italy over the hump, but not yet "finished"

US not yet at peak

•  I will assume, from looking at these and other country graphs, that the peak for the US CASES will will be reached on April 11, 2020.  But what about Daily Death peak??  I decided to look at the Daily Case and Death numbers to decide what is the delay between being tested positive and dying which likely are not the same day.  As you will see in the graphs below (looking closely), I see a 3-5 day difference between when the Case numbers begin to move above historical average and when the Deaths move above their average.

• Now I expanded the April 4th Daily Death actual for 11 more days using the stable growth rate of either 9.8% growth determined from the Global results, or 20.8% which is the US result.  This produced two possible peak numbers for Daily Deaths in the US.  (7 days to get the the CASE peak, and then 4 more days to reach the Death peak as we just determined the Death peak post dates the Case peak by 4 days.)  
• I can now produce the Total Cumulative Deaths at its peak the on April 15 by adding up all the daily deaths to this date.
• Finally, studying the shapes of those countries that have braved their storm, it is clear that there is a skew in the total distribution of deaths to the recovery side of the curve.  Therefore, I took the peak Total Deaths and multiplied it by 2.1.  Double it to get the other half of the recovery curve and added an additional 10% to account for the skew.  This produced at Total Death Count for the US of 

200,000

•  If you use the global growth rate on Daily Deaths, this number drops to 62,000.
•  On the other hand, if you use the US daily death growth rate, and assume that the daily death peak is 7 days, rather than 4, after the CASE peak, this pushes the total deaths to 375,000.
• As a test for reasonableness, I produced Total Deaths Rate per 10,000 POPULATION to compare a few countries.  (Remember, I am skeptical of  Infection Rate numbers as the Corona testing has been limited and inconsistent across countries.

• In summary, my 200,000 projection represents highest number of deaths per 10,000 people of all these countries.  The days to peak, growth rates and measurement accuracy are all contributors to the wide swings in the projected deaths.  And this doesn't even consider all the complicated medical and viral issues!  The Asian countries in this list are orders of magnitude smaller.  One is communist and one is not, but the general culture is the good of the whole is more important than the good of the individual.  This could be an important learning if these results hold.