Hacker News Reader

Comment by qiqitori | Hacker News Reader

While 2% is indeed the WHO's current official early estimate, it's still very early to give accurate numbers here.

https://www.worldometers.info/coronavirus/ Using the "closed cases" figures on this page (currently: 27892 closed cases, 25265 recovered, 2627 deaths), we get a fatality rate closer to 9.4%, but I've been monitoring this page for a while, and the fatality rate based on this calculation has improved a lot. For example, on 2020-02-10, the figures were 4376 closed cases, 3466 recovered, 910 deaths, resulting in a fatality rate of 20.8%. (https://web.archive.org/web/20200210100748/https://www.world... <3 archive.org)

View on HN

World-o-meter isn't the definitive source for that information. Does the big counter at the top include the large number of people who contracted the disease, but never progressed past a mild stage, and so never reported it to their doctor?

The WHO number is based on more than just the known cases vs deaths tally.

> Does the big counter at the top include the large number of people who contracted the disease, but never progressed past a mild stage, and so never reported it to their doctor?

The big counter is based on official stats reported by the various countries and is equal to the numbers published by the WHO. Obviously, nobody knows the number of people who had it but didn't report it. You could do random sampling and get an estimate for the particular population that you tested, but let's forget about that for a moment.

The WHO's fatality estimate may or may not turn out correct at some point for the entire outbreak, but sadly there is no denying that the number isn't correct for the people who were tested positive a couple weeks ago, and the WHO published their estimate a couple weeks ago too.

> You could do random sampling and get an estimate for the particular population that you tested, but let's forget about that for a moment.

If there was some benefit to the disaster that was the Diamond Princess, I guess it was that we have a pretty good idea about the range of disease symptoms. The various evacuees too - living in quarantine till they get the disease or don't.

> Obviously, nobody knows the number of people who had it but didn't report it.

Just because we don't know the exact number doesn't mean there aren't ways to estimate.

My point is that estimating a mortality rate involves a lot more than simply doing a single arithmetic operation.

I have been confused about this 2-3% figure since they started using it... I mean they are not factoring in the lag-time.

It seems like they just are taking all the new patients, and not calculating it correctly. I expect this number to grow.

This NIH representative assessed the fatality rate at more like 1% in an interview to PBS a few days ago: https://www.youtube.com/watch?v=a6-De7IAS1I

There are multiple forces at play. On the one hand, as you note, it takes time for people to die, which could initially bias the fatality rate downward. On the other hand, a lot of milder cases will be missed entirely, which would bias the rate upward. My understanding is that the latter factor is considered more significant in the early stages of an epidemic like this.

Reply on news.ycombinator.com