A new article in Nature has gathered statistics from several studies to come up with an estimate of the overall death rate from coronavirus (click on screenshot below to read it, pdf here). If you’re paywalled, a judicious request might work. I’ll put the latest estimates at the bottom, as you’ll need to read the preliminary information since these figures come with many caveats.
As the article notes, when you’re estimating fatality rates, the gold standard is called the “infection fatality rate” (IFR), which is the proportion of all infected people, including those who are asymptomatic or haven’t been tested, who will die from the disease at issue. You can imagine the difficulty of estimating this. While we can get an accurate handle on the fatality rate among those known to have the disease, that’s only a part of the statistic, and may either under- or over-estimate the IFR. Further, if you have antibodies against the virus, you may have recovered from an infection without knowing you’ve had it. Yet that data must also be incorporated into the IFR, and antibody testing is not the same thing as testing for the virus. (How many of you have been antibody tested?) One study from Germany showed that 15.5% of the people in a town that had an outbreak had coronavirus antibodies—five times the proportion of people known to have had coronavirus at the time. Not doing antibody testing would have drastically overestimated the IFR.
Another complication is that some countries don’t test postmortem, and, importantly, the fatality rate in different groups (age, ethnicity, class and wealth, comorbidities, access to healthcare) haven’t been compiled thoroughly. Of course, if you’re infected or in a group that doesn’t have the average IFR, you’ll won’t care that much about the overall rate—you’ll want to know your own chance of dying.
Why do we need these data? As Nature notes:
Getting the number right is important because it helps governments and individuals to determine appropriate responses. “Calculate too low an IFR, and a community could underreact, and be underprepared. Too high, and the overreaction could be at best expensive, and at worst [could] also add harms from the overuse of interventions like lockdowns,” says Hilda Bastian, who studies evidence-based medicine, and is a PhD candidate at Bond University in the Gold Coast, Australia.
The article outlines other complications, but there’s no need to go into them here. I’ll just add that Nature presents the rate of six studies from five countries, and there isn’t much variance among them, with the first study, using data taken from a cruise ship in which everyone was tested, gives the only estimate of the true IFR. But the sample (3,711 people) was small.
So here are the data at hand, and realize that there are problems with all of the studies. But it is interesting that they tend to converge on a value of 0.5% to 1%. (Of course, if you’re old like me, or have other medical issues, this will be an underestimate):
Some scientists impute the small scatter to “luck” (whatever they mean by that) or coincidence, and virtually none of the data have been published in peer-reviewed manuscripts. Finally, of course, we need to know the death rate for different groups, which will help in figuring out individual treatment, though for epidemiological purposes the IFR is necessary—if it’s from a random sample of people. (Nature cites one study from Switzerland estimating an overall IFR of 0.6%, but a tenfold higher rate of 5.6% for people 65 or older.
The lesson: so far across several populations, one’s chance of dying should they contract the disease is about 0.5% to 1%. But your mileage may vary (I have a lot of mileage and my figure would be higher), and it’s early days for these statistics.