What Does the Infection Fatality Rate Really Measure?

This paper studies what the infection fatality rate (IFR) really measures using the potential outcomes framework. I show that the IFR only reflects the outcome in one state. In contrast, popular causal parameters are all functions of the difference between outcomes in two states. I then demonstrate using a simple illustrative example that a disease that has no effect of the risk of dying can have a higher IFR than a disease that increases the risk of dying for everyone in the population. As a result, the IFR may fail to reflect the causal effect of a disease on the risk of dying and hence might not be a suitable measure of how deadly the disease is.