How deadly is the new coronavirus?

AS THE NUMBER of deaths among people infected by the new coronavirus mounts, it is tempting to divide that figure by the number of reported cases and conclude that the result is the fatality rate. Apply such maths to the world’s total of confirmed cases and deaths on March 11th and you get a fatality rate of 3.6%. But this figure, which epidemiologists call the “naive” case fatality rate, may be wrong in two different ways. First, many of the infections detected at this early stage of the epidemic are recent, so some will eventually result in deaths. That will push the fatality rate up. Second, many infections have not been spotted because testing for the virus has been patchy. Lots of mild cases of the disease have gone unnoticed. If all infections were actually counted, the result would be a bigger denominator. That would push the fatality rate down. As China began to trace infections more carefully, its fatality rate fell (see left-hand chart).

A conclusive measure of the denominator requires testing for antibodies against the virus in a large sample of people in a place which an outbreak has already swept through. Such studies are under way in China. In the meantime, researchers have estimated the fatality rate for covid-19 using a cohort of people for whom there is a full count of infections and deaths: passengers on the Diamond Princess cruise ship. A bungled quarantine on board led to nearly 700 cases of covid-19. Eight people have died so far. A working paper published on March 5th by Timothy Russell of the London School of Hygiene and Tropical Medicine and an international team of researchers estimates that the fatality rate among infected passengers will end up being 1.2%.