If you are infected by the coronavirus, i.e. if you contract covid-19, what is the probability of your dying from it? What is the death rate of this particular virus? On Tuesday March 3 the World Health Organization (WHO) stated that the global fatality rate was 3.4 percent. The figure changes daily, but if we take this to be a correct measure (it isn’t as I explain below), your own probability of dying from the disease will be higher than that if you are older and/or suffer from other health immune compromising conditions and lower if you are young and healthy, etc. Oh to be young and healthy again.
Yesterday, Friday March 6, the WHO reported 3,400 deaths worldwide and 100,000 confirmed cases worldwide. This is the basis of the 3.4% death rate. 3,400/100,000=0.034=3.4%. The true figure is probably lower, but it could also be higher. The numerator of this fraction–deaths–should be pretty certain. However, mistakes can be made in stating the cause of death. A covid-19 death could be missed and attributed to something else.
The biggest probable source of error is with the denominator: the number of cases. There is a high likelihood that many who have contracted the disease have not yet and may never be diagnosed to have it. It is likely that these missed cases are milder than those that are identified and thus have lower death rates. The U.S. has been particularly slow in testing for the coronavirus. The number of test kits needed to perform such tests are still very limited in the U.S. As of yesterday (Friday March 6), the U.S. had only tested 1,890 people of which about 10% were positive. By comparison South Korea has been testing about 10,000 people per day for weeks. To the extent that the actual number infected with the virus is larger, and probably much larger, than those identified, the denominator will be larger and the death rate lower. By comparison in the 2017-18 flu season in the U.S. 45,000,000 had the flu and 61,000 died.
But that is not the end of the challenges to an accurate figure for the death rate. The 17 deaths so far in the U.S. may, and probably does undercount the number who will die from those currently infected. If some of those already infected will subsequently die, the true death rate will be higher. The true figure will only be known sometime after the event. The following article provides a very good discussion of this issue. https://www.washingtonpost.com/health/coronavirus-mortality-rate/2020/03/06/b0c4cdfc-5efc-11ea-b014-4fafa866bb81_story.html
**************
Dr. Briones adds the following description of a more scientific approach to estimating the death rate.
There is the “TRUE” death rate and there is the “EXPECTED” death rate. The True death rate answers the question of how many people died from a particular disease event. The expected death rate represents the more computable number when people ask for the probability of dying from coronavirus infection.
TRUE death rates can only be computed after the “event” has finished and all people involved have been counted including those that survived the event and those that died from the event. For example, a pile up car accident involving, say 100 people, with 20 deaths would have a TRUE death rate of 20%. The TRUE death rate in this situation can be calculated accurately because the event has finished and is no longer evolving. No more people will be involved in this particular situation in the future.
In the present coronavirus epidemic, the TRUE death rate CANNOT be computed because the infection and disease continue to evolve and increase. It remains impossible to determine how many people actually have been infected with the Coronavirus as compared to those who have been tested and found to be positive for the virus. Also, most importantly, the TRUE death rate cannot be established because all those that have been tested positive have not finished the disease process. The TRUE death rate requires knowing the number of people that recovered/survived after coronavirus infection as opposed to those that died from the infection. Since the present situation of the Coronavirus epidemic continues to spread and infect more people, the result cannot be computed. This may be possible after the Coronavirus epidemic has been contained and new infections have been significantly minimized.
The better computation to answer the question of probable mortality rate from Coronavirus infection would be the EXPECTED death rate. In this computation, a randomly selected group of individuals from the pool of people who have tested positive from the Coronavirus are followed to the end of the disease process. (This approach assumes that this group is representative of the whole population infected by the Coronavirus.) So far, people infected with coronavirus either recover after two weeks or so or they die from the disease due to respiratory distress.
For example, 100 people are randomly selected from those positive for the Coronavirus. After a month or so, these 100 people would either have recovered or died from the infection. If 2 people died from this selected group, then the expected or probable death rate would be 2%. One important key point in this experiment is to follow the disease process to its conclusion of either recovery or death.
To increase the “Power” of the result of this experiment, researchers can further increase the randomly selected population from 100 to 1000 or even 100,000. But it is important that those selected to be part of the final computation (number of deaths divided by the total population) should have finished the disease process.
To increase the accuracy and power of the EXPECTED/PROBABLE death rate is to further categorize the selection by age group. For example: 20 to 40 years of age versus 50 to 70 years. Another refinement of the group selected can address older people with underlying conditions selected for the study. Further refinements of this population group can be made in order to accurately answer the question of: What is my (male vs. female, young versus old, no underlying disease versus one with underlying disease, etc) probability of dying from Coronavirus infection.
With already over 100,000 Coronavirus infections around the world, these numbers and computations can be done on a retrospective analysis of the various population groups.
******************
WC: The above approach should produce a more accurate estimate of the death rate if the death rate of those testing positive is representative of the death rate of those not tested. It is reasonable to assume that those who were not tested but had the infection and where thus not part of the population from which the test sample was drawn, had a milder reaction to the infection and thus probably a lower death rate (possibly even zero). Thus the “expected” death rate computed in the way explained by Ito above would be a maximum expected rate and the actual rate is likely to be lower.
Warren's space 
Perhaps something to consider: values for the regular flu are also just estimates.
“Total flu-related deaths during 2017-2018 was previously estimated to be 79,000, but the current estimate is 61,000.”
https://www.cdc.gov/flu/about/burden/2017-2018.htm
Trump says that “3.4 percent number is really a false number.” He thinks that 3.4 percent is too high because it overlooks unreported cases. On the other hand, 3.4 percent is too low because it compares deaths now to cases now instead of cases roughly two weeks ago out of which some people died.
We can get a better idea of the true death rate by adding known recoveries, 58,400, to a guess of unreported recoveries, say 43,000. That gives us 3600 deaths (at last count) per 58,400 + 43,000 + 3600 cases or 3.4%. So 3.4% is a good number to the extent you believe that 43,000 recoveries have gone unreported. Only if you think 43,000 undetected recoveries is too low, do you agree with Trump that 3.4% is too high.
I like to think that 43,000 is correct. That way 3.4% is a “true number” because two wrongs make a right.
Your logic is sounding like Trump’s. Lol
Actually, 43,000 is probably too low. WHO reports that “suspected cases” in China (for which the test was inconclusive) are 97% of confirmed cases. And that still omits the untested cases due to:
1. Insufficient testing supplies
2. Those w/ mild cases not seeking treatment.
3. Those avoiding being quarantined.
What effect, if any, did the 61,000 (43,000 ?) flu related deaths have on the stock market in 2017-18?
None. Today’s fall is correcting an equity price bubble and reflecting concerns over the KSA – Russia oil price war effect on fracking plus uncertainty over were the new virus might take us