Case Fatality Ratio (CFR)

The case fatality ratio is deaths divided by diagnosed cases: of the people known to have had the disease, what share died of it? It is a severity measure, and its long-standing name — case fatality rate — carries the same defect as the attack rate. The CDC's own Principles of Epidemiology flags it as a misnomer for the same reason: with no time in the denominator it is a proportion, not a rate. The preferred term is case fatality ratio, and "CFR" conveniently abbreviates either.

The number's real difficulty is not its name but its denominator, and every controversy about a published CFR is a denominator argument. Mild and asymptomatic infections are never diagnosed, so the case count is a subset of the infected — usually the sicker subset — and the ratio is biased upward, often by a lot. The infection fatality ratio, which uses estimated total infections instead, is the figure that answers "if I catch this, what is my risk", and it is typically several times smaller. Early in any outbreak the CFR is calculated from the hospitalised cases, because those are the only cases anyone has found yet.

The numerator has its own timing problem. Deaths lag diagnoses by days to weeks, so dividing today's cumulative deaths by today's cumulative cases understates the eventual ratio while an epidemic is growing, and overstates it while one is shrinking. During rapid growth the error is large. Cohort methods that follow a fixed group of cases to a resolved outcome avoid it and are correspondingly slower to report.

Finally, a CFR is a property of a disease, a population and a health system together, and never of the disease alone. Age structure moves it enormously for conditions whose severity climbs with age; so does access to critical care, and so does how hard a country looks for mild cases. Two honest CFRs from two countries can differ several-fold with no difference in the pathogen at all, and comparing them without age-standardising compares the populations rather than the disease.