COVID-19 Infection Fatality Rates

Claus O. Wilke, March 17, 2020

To see the code that produced this document, please click here.

Seasonal influenza fatality rates (from 1997 to 2007) were estimated by Jesse Bloom. He took all-cause influenza-associated mortality rates from Table 1 of Quandelacy et al. (2014) and assumed a 10% incidence rate in the population.

COVID-19 fatality rate data are taken from Table 1 from Ferguson et al. (2020). Importantly, these rates are corrected for infections that remain undetected (i.e., do not become cases). Actual case fatality rates are approximately a factor of 2 higher, see Verity et al. (2020), Table 1.

We now combine these two datasets and plot them. We use a log scale for the y axis because it will show a uniform increase in risk as a shift of one curve relative to the other.

We see that indeed the COVID-19 curve is systematically shifted relative to the seasonal influenza curve, by approximately a factor of 5-10. (The fatality rate estimates are too imprecise to try to quantify this more accurately from this type of a crude analysis.)

For comparison, we can also use a linear scale for the y axis. This representation emphasizes the high risk that COVID-19 poses for the elderly, but it hides the fact that for younger people COVID-19 is also much more risky than the seasonal influenza.