I’m still seeing a lot of confusion around mortality rate figures for Covid-19.
There are three figures worth understanding.
The first is the case fatality rate (CFR). It measures the proportion of cases of a disease that result in fatality. To calculate it, take the number of fatalities and divide it by the sum of the number of fatalities and the number of recoveries. It’s less useful in the middle of an outbreak because a lot of cases haven’t resolved in death or recovery yet.
The second is a (not great) approximation of the case fatality rate calculated by taking the number of deaths and dividing it by the number of cases reported. The problem with this is that there may be many cases that have been reported but which have not yet resolved. This means it incorporates a time delay that distorts it significantly if the rate of cases is not constant. A growing outbreak will undercount fatalities because there are fewer people who’ve had the disease long enough for it to resolve in death compared to those who have only been infected recently and thus have not died.
The third is the mortality rate, which is the proportion of people who die from the disease to those who get it, regardless of whether they are detected or not. This is the figure that matters if you’re trying to understand “if I get it, will I die”?
These figures are distinct and have different limitations. The first two are heavily dependent on how much effort is put into diagnosis; if you do a lot of testing, you find a lot more mild cases, and your CFR goes down. The last one is impossible to know for sure if you don’t have perfect testing.
Using data taken right now from worldometers.info, the following figures are these:
- The current case fatality rate is 6.6% (4983 deaths / 75377 closed cases).
- The poor approximation is 3.7% (4983 deaths / 134769 cases).
- The infected mortality is estimated to be around 1% .
So, if you’re seeing different figures in the media and wondering why, it’s because people are using different calculations and not being clear enough about where they’re coming from.
By way of comparison, the Spanish flu, which killed millions, had a CFR of 2.5% or possibly a bit more (wikipedia cites an article which gives it as >2.5% without any explanation). That makes the current CFR for Covid-19 something to be seriously worried about.
Worldometers has a page of notes that get into this in more detail than I can here.
 The best example I’ve seen substantiating this is the Diamond Princess. Everyone coming off that ship was tested, so its reasonable to claim that most if not all cases of the disease were identified, meaning that the CFR can be expected to be pretty close to the mortality rate. Of 696 cases identified, there have been 7 fatalities, almost exactly 1%.
 A quote in a Swiss Medical Journal describes this method thus: “At present, it is tempting to estimate the case fatality rate by dividing the number of known deaths by the number of confirmed cases. The resulting number, however, does not represent the true case fatality rate and might be off by orders of magnitude”.
 This 3.7% figure is related to the 3.4% figure that’s been thrown around. It appears to come from a quote by the head of the WHO, who said “Globally, about 3.4% of reported COVID-19 cases have died.” on March 3. Unfortunately, the exact meaning of his statement has been widely misinterpreted.