How Do I Make Sense of Coronavirus Risk Numbers?

By March 26, 2020 March 29th, 2020 Commentary

A lot of information is thrown around on supposed infection rates and death rates for the current coronavirus.  Here is a suggestion for how to think logically about all these numbers.

1) First, what percent of the population has had exposure to the virus. We don’t know what that is, but I assume a high percent of the population has been or is going to be exposed. But it clearly isn’t 100%. Testing for antibodies is the only way to try to partially determine that number, see below, and some studies are beginning.  But antibodies are only generated by people who are infected.  Not everyone who is exposed is infected.  If there isn’t an antibody specific to this variant, it would be hard to know if the person was exposed or not.  I don’t think there is a good way to determine exposure rates.

2) Next, what percent of the population has been infected, which means you were exposed and the virus actually got a toehold in your body. As I note above, not sure how accurately or by what means we can determine how many people have been exposed.  Infection rates may be more important in any event, and as best we can tell, that is actually low, only 20% on the Diamond Princess, apparently much lower in Wuhan, where you have to remember the virus ran unchecked for weeks in a densely populated metropolitan area of over ten million people, likely exposing a very large fraction of the population.  I have more information on this in a separate post.  Why it is low is an interesting question, it suggests some general coronavirus antibodies are present in most people and operating against the virus, or that there is some other protective factor at work for a large percentage of the population.  To ascertain infection rates, you would need large scale, randomized testing of the population for antibodies.  This number is the one which is most abused by politicians, along with death projections.

3) Then, what percent of the infected will develop an illness, which means you actually develop some symptom due to infection by the virus.  Again, as best we can tell from testing rates and other factors, a very large percent of those infected are asymptomatic, at least half and potentially as many as three-quarters of all who are infected.  Hence the concern about social distancing, good hygiene, etc. because if you are infected, you can be spreading the virus even if you are asymptomatic. To ascertain the true rate of asymptomatic persons,  you would need a large scale randomized testing study for presence of the virus, and you would need to find out how many people who had the virus had a symptom.  Since for most people, the illness is pretty mild, you would be relying on self-reporting, which can be unreliable and at this point you would assume many people with any of the typical symptoms are assuming it is due to coronavirus, when it often certainly isn’t.   Anyone who has tested positive and seeks medical care would also be included in that percent of symptomatic people.

4) Next, what percent become seriously ill, which you can define as someone who needs hospitalization.  This also gets a little tricky, because the people who are most likely to get infected and become seriously ill, generally have multiple serious comorbidities.  So is it the virus or the comorbidities that is causing the need for hospitalization.  But at least if you got your denominator right, see the note below, the calculation of the percent of people who get seriously ill should be more straightforward than some of the other calculations, because we have good information about diagnoses for hospitalizations.

5) What percent of the population dies, again not as straightforward as it seems.  Italy’s experience is illustrative.  Did these very sick patients who also got coronavirus die from coronavirus or the other diseases.  In Italy a review of death certificates determined that only a fraction of deaths attributed to coronavirus illness were actually due to it.  But if your definition of cause of death is good, percent of deaths is a straightforward calculation.

One final note is the denominator issue.  Best in my judgment to use population, which we do know with precision, so all the above numbers are expressed as percents of population or rates per million of population for example.  But no matter what is used as your denominator, you have to be clear about it.  If you are going to use deaths as a percent of infection, say that is what you are doing and acknowledge that no one knows what the true infection rate is in a large population like the US. Governor Walz and others are using denominators that are not population and not being clear about it.  Obviously any denominator other than total population makes the percent or rate look higher and scarier.  And any denominator other than population is useless at this point because of the issues mentioned in 1, 2 and 3.

Leave a Reply