Humanity's understanding of this disease is much more sophisticated than the two comments above this would imply.
Researchers are not stupidly looking at Diamond Princess and getting confused because the population is old.
Rather it's a powerful dataset because it enables us to quantify the asymptomatic people of all ages, because people were tested fairly exhaustively before being allowed leave, regardless of symptoms, and because we've had enough time for many of the cases to play out.
The diamond princess data revealed a large number of asymptomatic cases in China, the absence of which were skewing the detected cases in China much older.
Anyway, you don't need to think anything complicated about correcting for ages, etc
You can just look at the data from the Netherlands where they conclude they have 3% population infection, by looking at the random serological testing, and then look at the number of confirmed deaths they have, and do the very simple math.
It's very hard to do that and then say we're missing say 10x higher numbers of asymptomatic cases all over the place.
Anyone who actually knows this stuff would never say that it’s a matter of “simple math”, that’s absurd. Those people have not been saying that they’re sure about anything at this point, because they’re not.
There’s nothing to prove you wrong yet either, of course, but let’s see how your prediction does over the next few months: between 14,400 and 80,000 deaths in Ireland.
The simple math comment was about estimating IFR, given the sampling in the Netherlands, not about the whole domain.
Maybe you are saying the implied model is too crude. Perhaps, but there's a value in it's simplicity - folks can get a convincing estimate of the rough scale of the IFR, that this disease is serious, not the flu etc.
> let’s see how your prediction does over the next few months: between 14,400 and 80,000 deaths in Ireland.
That prediction, was based on the measures in place over a month ago (very few, which was the point).
We're since in lockdown almost 3 weeks.
However, I think it remains a reasonable prediction for an unmitigated epidemic.
To go back to the simple math: If you believe that 3% of a random sample of the Netherlands have had covid, and that the Netherlands has had 3k deaths, then if 100% were infected they might have 100k deaths (without modeling clustering, and healthcare saturation effects). Their population is 3x Ireland's.
>There’s nothing to prove you wrong yet either, of course,
If the serological sampling had shown instead that their 3k deaths had come from 30% of their population being infected, that would have proved the prediction wrong.
There's loads of subtlety which could improve that estimate (people take time to die, clusters, etc) but there is an increasing body of evidence that the early IFR estimates were not wrong by an order of magnitude.
You really think that if your numbers don't pan out, it will be because Ireland somehow suppressed this thing (put that "genie" back in the bottle as it were), and not simply because its IFR is (or became) lower than you assumed it would be?
It's just splendid how a lot of people will criticize an estimate of deaths assuming no mitigating measures, and then say "see? no problem!" when the outcome is much better after stopping most economic activity for a month.
When you look at countries where front was, it looks different. It is quite cherry picking to pick the best off countties. One of then had no war on its soil (and earned finacially overall).
I would caution about taking dutch death numbers as truth. We have an acknowledged deflated death number. Because it states only confirmed cases, whereas we have a large number of home deaths (elderly deciding not to go through ICU traima) that are not tested and hence do not count.
Researchers are not stupidly looking at Diamond Princess and getting confused because the population is old.
Rather it's a powerful dataset because it enables us to quantify the asymptomatic people of all ages, because people were tested fairly exhaustively before being allowed leave, regardless of symptoms, and because we've had enough time for many of the cases to play out.
The diamond princess data revealed a large number of asymptomatic cases in China, the absence of which were skewing the detected cases in China much older.
Anyway, you don't need to think anything complicated about correcting for ages, etc
You can just look at the data from the Netherlands where they conclude they have 3% population infection, by looking at the random serological testing, and then look at the number of confirmed deaths they have, and do the very simple math.
It's very hard to do that and then say we're missing say 10x higher numbers of asymptomatic cases all over the place.
This blog post I wrote a while back is Ireland specific but has figures on CFR and IFR I think are still valid, and references to the academic articles I got them from: https://medium.com/@fergal.reid/predicting-the-impact-of-cor...