Monday, May 4, 2020

What do all these numbers mean?!

Putting all these statistics into context
(Brace yourself, I'm about to do math...  There are some numbers "out there" these days, but not much context.  It might be easier to understand them by comparing to something that people are familiar with:  The ones that we can be most certain about are historical flu data.)

Same as, worse than, or better than the flu?
Publicly available information about past flu seasons and their average values are what I'm going to use.  On average, the flu season lasts about 13 weeks (or 91 days).  It peaks some time between December and February, so it crosses over the new year and is counted like school years ("2010-2011").   The average number of deaths among the 2010-2011 season to the 2016-2017 season are 34,571 deaths per flu season.

The assumption I'm making here is that we pack all the deaths into that season, nothing falls outside of it.  (There are usually outliers, but I'm just spit-balling here.  Please bear with me.)

34,571 flu deaths in an average season / 91 days = 379.9 deaths per day in an average flu season—round up to 380 because we can't have 0.9 of a person. 
380 deaths per day, spread out evenly across 50 states:  380/50 = 7.6 deaths per state per day.  

If you have more than 8 people dying of Covid-19 in your state every day, this is worse than the flu.   That's averages, some states are less populous; some years are more severe (11 deaths per state per day in a bad year).  

The Covid picture right now - Assuming we have no more deaths from Covid, the deaths are spread out evenly over the 57 days since the first death, and the deaths are averaged over 50 states, we get the following. 

67682 deaths this Covid season/ 57 days of death = 1187.4 deaths per day.  (rounding down to 1187)
1187 deaths per day / 50 states = 23.74 deaths per state per day, right now.  (May the 4th be with you.) 

Conclusion:  This is not "just the flu".  If people stopped dying right now, it would still be twice as bad as the worst flu epidemic in the last 10 years. 


What about this death rate?
We can't get a 100% accurate picture of who had the flu, so estimation goes on based on surveys and past history.

Figure 1: Illustration of Influenza Burden Estimates Model https://www.cdc.gov/flu/about/burden/how-cdc-estimates.htm

"The numbers of influenza illnesses were estimated from hospitalizations based on how many illnesses there are for every hospitalization"  (https://www.cdc.gov/flu/about/burden/how-cdc-estimates.htm#References #5)

O.K.  For 2010-2017, the average number of symptomatic illnesses per season is 25,328,000.  Over that time, the average number of deaths was 34,571.
(34,571/25,328,000) * 100 = Estimated 0.13% death rate for Influenza infections.

One of the reasons the Death Rate for Covid-19 is so high is because we don't know the number of infected. Only the number of confirmed cases.  There hasn't been any past episodes of this disease to get a handle on how many people don't seek medical care.

We do know how many people are hospitalized and how many people have died with more accuracy. If we look at just deaths to hospitalizations, the flu looks like this:
(34,571 / 388,571) * 100 = Estimated 8.8% death of those that are hospitalized.

So how many people in the US have been hospitalized for Covid-19?    About 126,298 according to Johns Hopkins data as of 5/4/2020, and 67,795 dead. ( https://coronavirus.jhu.edu/map.html )
(67,795 / 126,298) * 100 = Estimated 53.7%  death of those that are hospitalized.

Now we can compare apples to apples.  8.8% hospitalized death rate for flu, 53.7% hospitalized death rate for Covid-19.

Conclusion:  This is about 6 times worse than the flu.  i.e. If you get Covid and you end up in the hospital you got a little less than a 50% chance of going home just fine.


How many people are actually going to get this?
Let me back up for a second now that I got the 'more concrete' numbers out of the way. ('Still some wiggle room based on reporting to health departments.)

Average number of flu illnesses per year = 25,328,000

Statisticians are trying to get a handle now on how many people are sick, but not officially confirmed as having Covid.  "As of March 31, Vollmer and Bommer calculate, confirmed cases represented just 3.5 percent of infections in Italy, 2.6 percent in France, 1.7 percent in Spain, 1.6 percent in the United States, and 1.2 percent in the U.K. In other words, the true number of infections was between 29 and 83 times as high as the official tallies in those countries." [ https://reason.com/2020/04/12/official-covid-19-numbers-represent-just-6-of-total-infections-a-new-analysis-suggests/ ]

Now we're going out on a limb at making more estimates.  So don't take my word for it.

If these guys/gals are right then today's number of confirmed case represent  (1,161,804/ 1.6) *100 = 72,612750 suspected cases in the US.

The US has  328,200,000 people.    That's a suspected 22.1% of the population that is infected . . . seems a little high, but this is just an estimate.

If we assume all other things are consistent, and if we all just ran out and got infected on purpose to push that up to 100% infected, we could have the number of hospitalizations up to 571,484  (126,298 / 22.1 = 5714 hospitalized per 1% of the population;  5714 *100 if 100% of the population were infected).  And we could push the number of deaths up to  306,315.  (571,484 * .536)

People are "naive", immunologically, because there have been no previous Covid-19 out breaks and there is no vaccine.  So people will get it if they are exposed enough.

Worst case scenario, if confirmed positives only represent 1.6% on the infected, 238,521 deaths on top of what we have right now if we stop all the social distancing measures.

If the above Statisticians are wrong, and testing is actually more complete, say 10% of actual cases are confirmed, then the numbers change a lot.  Only 3.5% of the population is infected so far.  The number of hospitalized is closer to 3,608,514.  And, the number of dead is closer to 1,937,772.

So . . . how good is our testing Really?  How many people have really gotten this?  


~NG

No comments:

Post a Comment