Hacker News Reader

The "official" 200k estimate (https://www.medrxiv.org/content/10.1101/2020.03.27.20043752v...) assumes that China isn't lying about their numbers, that Italy's official numbers are correct (Even though the death rate in some places there is 5-10x higher than normal), that the US will be able to implement "lockdowns" as effectively as China did, and that the rate of decline of new cases of COVID-19 will match the rate of increase before the peak.

I'm not sure any of those assumptions are warranted.

The thing missing here is that you're looking at deaths without looking at number of cases of each. The flu causes the number of deaths it does because it infects tens of millions of people a year[0]. The death rate is an order of magnitude lower than with COVID-19 across every single age bracket. [2]

So when people say things like:

> "I'm not worried about getting this illness"

They are likely not considering that even the younger are taking as much as a 1% risk of their life, which is incredibly high. Depending on factors like the load on the medical community at the time and others, it can get as high as 5% or more quite easily as we have seen around the world. Currently using only napkin math from the numbers here[1], the death rate is currently at just under 4% over all age groups in the US.

Additionally, it is not considering the danger they are putting others in by getting it. With the flu at a death rate of 0.2% or lower, causing deaths by infecting people is a very small risk. Passing it to one person here significantly ups the chances of you causing someone else to die.

Finally, flu deaths almost exclusively happen in the immunocompromised and elderly, while COVID-19 affects younger people a LOT more.[2]

[0] https://www.cdc.gov/flu/about/keyfacts.htm

[1] https://www.nytimes.com/interactive/2020/us/coronavirus-us-c...

[2] https://i.insider.com/5e81f6460c2a6261b1771b05?width=600&for...

> They are likely not considering that even the younger are taking as much as a 1% risk of their life

I know the numbers are still super fuzzy and it depends on who you ask, but at least according to Wikipedia, the US fatality rate for ages 20-44 is 0.1-0.2%, not near 1% (though certainly at-risk individuals would be higher than the average...)

https://en.wikipedia.org/wiki/Coronavirus_disease_2019#Progn...

That "as much as" was attempting to account for overloaded medical conditions which will raise the rate, though for 20-44 it'd be unclear by how much. If we go by that table using a 2.3% death rate, that'd already be .3% using a naive literal scaling to the death rate for that age group for the US currently. But yeah, fuzzy numbers and napkin math warning is certainly warranted, I'm nowhere near an expert on this.

As mentioned, the flu death rate of 0.1% is overall as well - for younger people it is also significantly lower, so the increased risk still holds for younger folks. Source [2] shows that well.

> They are likely not considering that even the younger are taking as much as a 1% risk of their life, which is incredibly high. Depending on factors like the load on the medical community at the time and others, it can get as high as 5% or more quite easily as we have seen around the world. Currently using only napkin math from the numbers here[1], the death rate is currently at just under 4% over all age groups in the US.

Those are pretty high numbers. Using a Lancet study (https://www.thelancet.com/journals/laninf/article/PIIS1473-3...), IFR is just under 0.1% for a 30 something year old. Hospitalization rates are at 3.5% (2-7% confidence), so a death rate (for 30 year olds) even under a collapsing medical systems for younger people is highly unlikely to get close to 5%.

All said, a 0.1% death rate is still pretty dangerous and catching the disease is something like raising your risk of death by 50% in a given year. Lockdowns are still justified under those numbers.

> Finally, flu deaths almost exclusively happen in the immunocompromised and elderly, while COVID-19 affects younger people a LOT more.

Well, and old people even more. Are the relative risk ratios actually different for covid vs. flu - or are we just seeing the effect of a disease that is 7xish a really bad flu season?

(For young children actually, I believe covid outcomes are better than flu.)

> Lockdowns are still justified under those numbers.

And what if the effects of an extended lockdown is a tanked ecomony with a 3x increase in suicides for the next decade? Is it still justified? Lockdowns are only justified if the only variable you are optimizing for is "number of covid-19 deaths". Yet there are hundreds of other variables that we are ignoring in our quest to minimize that one variable. Impossible to tell now, but tanking the economy to flatten the curve may end up indirectly killing even more people in the long run.

Fair point. In some sense, I share your feelings a bit (in a lighter sense). The park closings some Bay Area counties are doing (e.g. San Mateo) feel not justified; I'm also not sure why the SIP was extended all the way into May given the existing downward trajectory. [Looking at both Seattle and Iceland, you don't need full-on SIP to keep R at 1]

> All said, a 0.1% death rate is still pretty dangerous and catching the disease is something like raising your risk of death by 50% in a given year. Lockdowns are still justified under those numbers.

0.1% of the rest of my life is ~3 weeks. The lockdown has already cost me more of my life than a confirmed case of COVID-19 would (on average).

And the math doesn't get much better the older you get, an 85 year old has the same chance of dying of natural causes during a 12 month lockdown as he has from actually contracting COVID-19.

That is an interesting point and I'm not sure how public health officials are treating it. I'm in tech WFH so I don't see even half my life being lost from the lockdown (maybe 20%), but YMMV.

Those % you are citing are highly misleading. Because they are % of tested positive cases. There's likely an order of magnitude more actual positive cases which have not been captured.

I simply do not believe those figures at all.

How can you even claim an "average death rate" when you don't even know how many people were infected, had mild symptoms, and recovered without reporting it to anyone?

Of course the numbers will look scary when you only count hospitalized people as infected.

> How can you even claim an "average death rate" when you don't even know how many people were infected, had mild symptoms, and recovered without reporting it to anyone?

We can either throw our hands up and give up or we can do the best with what we have. So, what percentage of cases do you think are unconfirmed? You can adjust my math with that and see what those rates are. Those numbers still stay scary unless you assume a 90% unconfirmed rate, at which it equalizes for young people compared to the flu but is still 10x or more for some age groups. The point being, even accounting for underreporting this is still much more deadly and allowing it to spread like the flu would result in many more deaths.

I haven't seen any reliable numbers given about the underreporting rate, so if you have seen any that would be interesting!

See the report by Imperial: https://www.imperial.ac.uk/media/imperial-college/medicine/m...

Looks like Spain had 6 million infections but only 140 thousand cases. So the actual uncomfirmed rate could be 97%.

You’re touching on what this whole thing is.

It’s rushed and woefully incomplete data enabling media companies to write apocalyptic headlines without exposing themselves to any legal liability. They have a fiduciary duty to their shareholders, so that’s what they’re doing.

I think to most people with a background in statistics it’s very easy to tell how bad the data is at a glance. It’s downright blatant. They don’t have a platform though, because in this media climate your platform is only as big as your level of sensationalism and hysteria.

There are multiple factors skewing the data toward coronavirus appearing more deadly than it actually is. It’s new and sudden, so there wasn’t a lot of infrastructure in place for testing. It’s causing a rush on hospitals. Combine those and you have one massive shortage of testing. You obviously want to care for the most critically ill first, so globally, only the most sick get tested, and only the cases most likely to be fatal are even represented in these numbers everyone is seeing.

Then they take wildly differing medical, economic and political environments all over the world, and merge all the data as if it’s equal. In many cases, in addition to the inadequate testing, that data is only as accurate as their governing bodies want it to be.

Do we really believe there are 5x as many cases in the US as in China?

Do we really believe there were only 5,000 cases in India?

Do we really believe the virus is 10x as likely to be fatal in Italy as compared to Germany?

Take the “confirmed cases” number and just add a zero, and you might be closer to the reality of what’s going on out there.

People are simply blind to the difference between “confirmed cases” and “cases”.

It’s very unfortunate that this nomenclature has a natural tendency to make “confirmed cases” feel like it must be more accurate, and also unfortunate that the term “cases” could only ever be an approximation. But those labels were chosen carefully and for a good reason, and they represent very, very different things.

The low 200k fatality estimate probably factors in the that the curve is flattening.

The current estimate is that 300k people are infected in the US. If the number of people infected by Covid-19 doubles every week, then in 10 weeks there will be 300 million infected in the US. Assuming 1% mortality puts you at 3 million dead.

Doubling each week is not how it happens... as a higher percentage of people have or have had the virus, the growth rate decreases because there are less people available to infect. That is why infection percentage is an s shaped curve.

I already baked that in by starting with a low infection rate (2x per week instead of 2x every 3 days).

Yes, but 90% is an unrealistic infection number no matter what you start with. I think most experts had it at around 60-70% if we do nothing.

That is not acceptable, either, given the fatality rate and the strain it would put on hospitals... but I think it is important we try to be accurate.

It's an s shaped curve, but we don't know the upper magnitude of that curve. A large percentage of the population would have to get infected in order to run the COVID out of new recruits. If it tops out at 50% of the population, that's a disaster.

Forcing the doubling rate to decrease short of hitting that point requires something else to happen. South Korea clearly did something other than running itself out of cannon fodder.

> as a higher percentage of people have or have had the virus, the growth rate decreases because there are less people available to infect

That is what you would call "herd immunity" and no, current numbers of recovered patients are not even close to have it. It doesn't happen like magic like some politician wants to make people believe, and current estimations for COVID _herd immunity threshold_ are 29%-74% of the community [1]. Make a count of that on the US population.

[1] https://en.wikipedia.org/wiki/Herd_immunity#Mechanics

Sure, but the comment I was replying to was suggesting that in 10 weeks, 300 million Americans would be infected... that is 91% of the population of the United States. We aren't going to hit that number.

The growth is not going to be that much, sure, but because we slow it down with isolation, not because the immune people stop it from spreading.

Yes... but even if we did nothing, we are unlikely to hit 90% infected... 60 to 70% yes, but not 90%.

I am not saying that is ok, or that we should just let it run its course... just that saying we will hit 90% infection is unrealistic.

One issue to consider is that the COVID wipes out our ability to treat serious flu cases by swamping the hospitals.

>Some of the clips show in the video referenced in the article actually seem pretty "sane" to me - the one guy saying "I'm not worried about getting this illness", another saying for most people it will seem just like the flu. Those two statements, as far as we know, are not that outlandish, right? I'm certainly not worried about myself, and for many people it does seem to be mild or asymptomatic?

The flu season is 6 months for ~ 40k deaths.

We are at 10k deaths of covid in the USA. Depending on how you measure the 'season' it could be as little as a month in. If we keep with these numbers it will surely pass the flu death toll for the same time period.

Now also consider that unlike flu, we have no vaccine and the virus kills people in the prime of their life just as much as the elderly. We also have no idea on the way this virus works. We are making assumptions that it will work like a normal flu virus. There is anecdotal evidence it resides in people even after they have 'recovered', which isn't the same as a normal flu or cold. It is certainly something to be concerned with.

> Now also consider that unlike flu, we have no vaccine and the virus kills people in the prime of their life just as much as the elderly.

Not with the same rates. This should be kept into account, because when lockdowns will be lifted (you will have to at some point, even if it's just to prevent people from becoming insane), you won't have a risk of 0, so letting the lowest-risk population out first is a goood bet.

> Now also consider that unlike flu, we have no vaccine

But there are drugs in development and in trial. None will work miracles, some will not work at all, some will have marginal impact, some will have a little more impact. Those are the best bets at this point in time.

Waiting out for a vaccine, like many governments are hoping for, is irrational IMO, because there's no guarantee that a working one will be found - and you can't lock up your population for 18 months, even if you don't factor the economy in, because people will be destroyed psychologically.