Doctors are being instructed to report all deaths associated with COVID-19 as caused by the virus. This method clearly overstates the impact of the virus because there are frequently other diseases present that contribute to the death.
Deaths caused by COVID-19 are overstated, but by how much? We can make an estimate using the data from Italy reported on Bloomberg.
There is a barrier to making such an estimate, which is that other illnesses are present in most cases. Suppose a person has 3 such illnesses and is attacked by COVID-19, with death resulting. Which of the four was the cause, at the margin? There is no way to decide among the 4 illnesses. If we start with 3 being present and the patient alive, then COVID-19 seems marginal and “the” cause. But we equally well could imagine a healthy person without these 3 illnesses getting COVID-19 and surviving. In that case, adding the 3 illnesses to the healthy person appears to be the cause of death. We simply cannot choose any one illness, COVID-19 included, as the cause of death. All 4 causes were present.
Because of this ambiguity, we proceed as follows. We are told that 48.5 percent of Italy’s deaths in which COVID-19 is a factor were of persons with 3 or more illnesses already present. Suppose there are 3 exactly and not more. (This makes the calculation if anything overstate the importance of COVID-19.) Add COVID-19 to the 3 that are present to get 4 illnesses. Divide 48.5 by 4 to get 12.125. Allocate that number to an estimated COVID-19 caused death total.
Repeat but for cases in which the patient had 2 other illnesses. These are 25.6 percent of the Italian total. Thus, divide 25.6 by 3 (which includes the added COVID-19) to get an increment to the total attributable to COVID-19 of 8.53 percent.
Repeat but for cases in which the patient had 1 other illness, which were 25.1 percent of the total. This gives 25.1/2 = 12.55 percent to add to the COVID-19 sum.
Finally, in the Italian case, there are 0.8 percent of cases in which no other illness was present. Add in the full 0.8 percent.
The sum is (12.125 + 8.53 + 12.55 + 0.8) = 34 percent attributable by this method to COVID-19. This means that if there were 1,000 deaths in total being said to be caused by COVID-19, a more realistic estimate of the impact of the virus in causing death is 340, which is a 2/3 reduction in attribution.
This reduced number is what may now be compared to the effects of other well-known causes of death.
Ignoring the presence of other illnesses hugely inflates the COVID-19 death count. If officials then focus on this inflated number in making decisions, their decisions will be enormously biased.
2:55 pm on April 12, 2020
