You Can’t Pay Any Attention to Infection or Case Fatality Rates

By May 10, 2020 Commentary

People use cases or infections of coronavirus interchangeably, and you will hear references to the case fatality rate or infection fatality rate.  That means the number of deaths that occurred for a certain number of cases or infections.  People look at that number as an epidemic proceeds to estimate what the end toll of an epidemic might be when it has worked its way through an entire population.  But it should be obvious to anyone that early on in an epidemic you may not have a good idea of how many cases there actually are, and that the most serious cases are the ones that will come to your attention.  We find ourselves in the current situation because policymakers relied on one absurdly high case fatality rate projected by one group, based on Chinese data.  That model, and the lead modeler, Neil Ferguson, have now been disgraced.  But we keep making the same mistakes in regard to case fatality rates.

Let me illustrate with a couple of examples.  Imagine an epidemic that created serious illness in half the people who were infected.  It hits your country, it infects 100,000 people.  50,000 show up at the doctor and need some level of treatment, including hospitalization.  1,000 die.  Testing has revealed that some people were infected but didn’t have serious illness, but you don’t know what percent that is.  So you think your case fatality rate is a bad 2%, maybe lower depending on how many unrevealed infections there are.  If you extrapolate that to the population of the country that would be exposed without intervention, maybe 70%-80%, you project millions of fatalities.  In reality your infection fatality rate was 1%, still not good at all, but not as bad as you believe.

Now imagine an epidemic in which the infectious agent only makes 10% of those infected incur a serious enough illness to seek treatment.  100,000 are infected, but only 10,000 seek treatment.  1000 die.  Now you think you have a disaster on your hands, a 10% case fatality rate, but you really have the same 1% case fatality rate.  These examples show how important it is not to pay any attention to case fatality rates unless and until you have a real sense of the number of cases.  And it is why as the coronavirus epidemic proceeds in this country and around the world we see the case fatality rate steadily falling, because testing is expanding greatly and random testing and antibody studies are beginning to reveal the true extent of infections.  And in fact there most likely are multiples of coronavirus cases compared to what positive test results show.  What can be more informative than case fatality rates, even now, is what percent of those serious illnesses, the one that need hospitalization, end up dying.  All the serious illnesses are likely to come to your attention and you want to understand likely outcomes of those hospitalizations.  But you can’t extrapolate those outcomes to the full population, without understanding the percent of illnesses that are serious.

My example shows an epidemic that is far worse than the current coronavirus epidemic.  The true case fatality rate for this one is going to be far lower that even .5%, it may even be much lower than .1%.  And, as I keep demonstrating, what is really significant about this epidemic is its dual nature.  For the general population, this is very rarely a serious illness.  The fatality rate in the general population is going to be far under a tenth of a percent.  For the infirm elderly, and a few other at-risk populations, it will be far higher, hundreds of times.  In such a bifurcated epidemic, you can’t let a general case fatality rate guide policy.

Leave a Reply