Looking at this morning’s statistics for the United States, I found that we had approximately 90K deaths and about 1.5M total cases, the latter figure including all cases ever–continuing cases as well as cases that had an outcome. Dividing 90K into 1.5M gives the result of 6%, which is higher than the 2-4% usually cited, but not massively so.
On the other hand, if we take the total number of resolved cases, that is 90K deaths added to 346K recoveries, and dividethe number of deaths into that total, then we get nearly 26%, which seems rather alarming.
What am I missing here? Every existing case of Covid-19 will, presumably, eventually end up in either the “Deaths” or the “Recovered” column, and the numbers so far are frightening.
Or is it the source of the statistics that I should be questioning?
(This website is updated continually so you won’t see my figures, but the proportions should stay about the same.)
I’ve been watching that same ratio you saw since the very early stages. I think Spain at one time was 55/45 on surviving/dying in the resolved cases. The more you can test the gen pop the more you’re going to find non-fatal cases that bring down the percentages though.
I’m not a medical professional but my guess is that there are four groups of people.
A) They caught it and they’re really sick. They report to a hospital.
B) They caught it and they stay at home.
C) They caught it and they’re asymptomatic.
D) They didn’t catch it.
In the early stages when tests were super hard to come by, group A got tested but not the others. But they were told to stay away unless they’re in really bad shape. By the time they get there the success rate is poor.
Group B may be tested now for antibodies? Or are officials conserving their tests, counting group B as recoveries?
Groups C and D may be tested if there’s a real emergency or if tests are plentiful.
Mortality rate is not the same as case fatality rate (CFR). Calculating the more accurate mortality rate takes time after the epidemic/pandemic is over. You cited case fatality rate. The rest of this post uses CFR.
The two-four percent case fatality rate (CFR) comes from a February 2020 initial medical study involving China-only victims. An updated study in March 2020 upped the CFR to 7.2 percent (Italy only). The articles does state, “the overall older age distribution in Italy relative to that in China may explain, in part, the higher average case-fatality rate in Italy.”
The actual CFR has been climbing since then as more and more data arrives.
Eventually we would like to know the infection fatality rate IFR. But that can’t really be known until we are past the epidemic and the dust has settled enough to know that those who are going to die have died, and those that are going to get infected have been infected. Then we need to estimate those numbers. OTOH, we are getting better numbers, through things like extensive random antibody testing. Assuming the antibody tests are now actually useful enough (which they clearly were not all that log ago) there are some number arriving that are suggestive of IFR. Spain did a test of 70,000 random people, and got an infection rate of about 5% of the population, which leads to a depressing estimate of IFR of 1.16%. Other studies suggest higher infection rates, and maybe lower IFR. But you would probably be putting money on the eventual worldwide IFR being in the range 0.5 to 1.5%. Of course there is deeper analysis to be done. Population demographics clearly matters. Spain has an ageing population. You might see quite substantial overall IFR changes comparing different groups purely as a result of nothing more than age distribution. Even that is open to deeper analysis, we have clear questions about racial factors, socio-economic, and so on.
I suspect comparisons of IFR with the Spanish Flu are not going to be easy. We didn’t have the testing available back then, so numbers are never going to be comparing like with like. Even CFR is subject to some variability.
A compounding factor is the nature of medical care. In an overwhelmed health system, we assume many who would not otherwise have died, will die. That means all death rates are subject to matters outside of just characterising the disease. We may also discover advances in effective management of serious cases that reduce the death rate in serious cases. There is certainly suggestion that that is already happening. Who knows, maybe a simple regime of flipping people prone, CPAP, vitamin-D, and being ready with steroids the moment things are looking bad can claw back a significant number who would previously have been on a ventilator with a grim outlook. Smarter people than me are working on this 7x24. Which is why IFR can only be calculated when the dust settles.
Except the statistics for the Spanish flu are guesses for the most part and we have not measured the Covid19 accurately enough to even produced educated guesses.
So the statement really has little significance, it could be much greater or much less.
This is the problem when you analyze information with grossly incomplete data, you go places that aren’t necessarily there.
Another issue with comparisons to the Spanish flu is that there were no antibiotics back then. Patients succumbing to bacterial secondary infections, especially pneumonia, were a big part of the deaths. That is something we know we can largely prevent now. So CFRs will be different. Quite possibly significantly so.
Even for exactly the same disease 100 years of medical progress is going to mean significant differences in outcomes and thus CFR. In poorer countries the gap may be smaller, but still significant. Poor does not mean stupid or incompetent. Sometimes one feels it is the opposite.