Hacker News Reader

Comment by arcticbull | Hacker News Reader

Funny you should mention that. You have approximately a 1% lifetime risk of dying from a car crash in the US [1], and your risk of dying from a COVID infection is 0.37% per the latest study from Germany (and it overwhelmingly only kills the old and sick). Given you develop immunity after (no evidence you don't yet), that represents a lifetime risk too. Which means, your lifetime risk of dying in a car accident remains about 3X higher than from dying of a COVID infection.

Lifetime risk is a better way of comparing the two because (a) both of them represent lifetime risks and (b) COVID is front-loaded, spreading furiously through the population so your daily risk today is higher than your daily risk of a dying in a car crash, but your lifetime risk remains much lower.

It's also worth calling out the flu has killed 2X as many people as COVID has to date this flu season. A bad flu season in the US kills about 60K people each year.

[1] https://www.iii.org/fact-statistic/facts-statistics-mortalit...

View on HN

Is the german study of germany? Because the ratio of recoveries to deaths is way higher in germany than the united states. The risk of covid varries a lot by geographic region and usa seems like one of the worst.

> It's also worth calling out the flu has killed 2X as many people as COVID has to date this flu season. A bad flu season in the US kills about 60K people each year.

So far. The flu season started before the covid season so that's a bit of an apples and oranges as you are comparing different time scales. The bigger issue though is its unknown how big covid could get if it balloons. Like its mostly on par with flu now, but its not over yet and that's with the taking of extreme messures to prevent transmission (which we dont do to prevent flu). Measuring potential risk in terms of what's happened so far when we haven't even seen what a complete season of covid look like, feels premature. Jumping off a bridge also looks safe if you base your predictions on the lack of all the bad things that happened before you hit the ground.

> Because the ratio of recoveries to deaths is way higher in germany than the united states. The risk of covid varries a lot by geographic region and usa seems like one of the worst.

You'd need to be a lot more specific. Which state are you referring to? Most US states are safer on Covid mortality than Germany. California + Texas + Florida (the three largest by population) combined have the same population as Germany with half the per capita death rate.

It would be like combining Italy, Spain and Greece and pretending they're all seeing the same situation.

Germany has a per 100k mortality rate of 5 for Covid.

Texas has a per 100k mortality rate of 1.5. California is 2.7. Florida is 3.6.

Germany has 89m people with 4,538 deaths. Texas has 29m people with a mere 453 deaths.

California + Texas + Florida = 89m people, with 2,290 deaths.

More large states? Ohio is 3.9. Virginia is 3. Arizona is 2.4. North Carolina is 1.5.

So as you can see, most of the US in fact doesn't look like New York, New Jersey, Michigan. Just like most of Europe doesn't look like Spain.

More countries in Europe for comparison (per 100k rate):

Belgium 47, Spain 44, Italy 38, France 29, Britain 23, Netherlands 21, Switzerland 16, Sweden 15, Ireland 11.6, Portugal 6.7, Denmark 6, Austria 5, Norway 3, Finland 1.6.

The three largest US states combined have a rate of 2.5.

Seems to me that according our current knowledge, all those numbers will end up at roughly the same number as time goes on. (Assuming you're counting the same thing.)

Lower, if your population is young and healthy. Much lower, if you miraculously stop community spread until medication is available.

Higher, if the disease spreads too fast and hospitals are overloaded, or if your population is old or unhealthy.

> your risk of dying from a COVID infection is 0.37%

I'm 55 so I'm pretty sure my risk is higher than that. Also, the covid risk is highly front-loaded. If I'm gong to die from covid, I'm likely to do so now whereas if I'm going to die in a car crash that might happen now or it might happen later. I don't really care about dying per se. Sooner or later it's going to happen. What I care about is dying sooner and more painfully than I have to. Drowning on my own bodily fluids doesn't sound like a pleasant way to go.

Your risk of dying is, accepting the 0.37% for the sake of argument, 0.37% if you're infected. The actual chance of anyone, even of someone in a more dangerous demographic, of dying of Covid in their lifetime is significantly lower than that.

Additionally, I believe the 1% chance of dying in a car crash is over the course of a lifetime. A 55 year old has likely lived well over half their life and has also aged past a significant chunk of the riskiest time to be in a car (likely 16-30).

> Your risk of dying is, accepting the 0.37% for the sake of argument, 0.37% if you're infected. The actual chance of anyone, even of someone in a more dangerous demographic, of dying of Covid in their lifetime is significantly lower than that.

I personally believe that given the R0 and the large number of asymptomatic cases, there's no putting this genie back in the bag, and I agree with prominent epidemiologists that we're all going to get it one way or the other -- because nobody's going to stay inside the 18 months it'll take to develop a widespread vaccine. That's why I support Sweden's model. [1] So I just rounded up to 100% and projected out.

> Additionally, I believe the 1% chance of dying in a car crash is over the course of a lifetime. A 55 year old has likely lived well over half their life and has also aged past a significant chunk of the riskiest time to be in a car (likely 16-30).

Good point! I wonder if it's a bimodal distribution. I assume older folks can't get out of the way fast enough as pedestrians and are more likely to get into accidents when driving.

[1] https://thehill.com/changing-america/well-being/prevention-c...

You support Sweden's model of ignoring the issue? Not only do they still have a high exponential growth rate, as well as exponentially increasing death counts, but their CFR is atrocious. Also the link in your footnote is extremely outdated, and doesn't mention anything about Sweden.

Sweden has actually leveled off and had flat new case growth over the last week [1], same as in Canada where they have shut the world down (in terms of flatness not numbers). Further CFR is almost exclusively based on age of acquisition and the extent to which hospitals are available. In Sweden hospitals are available. The disease is exactly the same everywhere so it’s much more to do with demographics.

I mean they’ve had 15000 recorded cases (of course since they’re letting the disease roam free without a general population testing regimen that’s going to be a minuscule fraction of the denominator which they expect to reach 7 million ish) and 1500 deaths. Crude CFR is 10%.

Bloomberg and the government as of 10 hours ago claim their strategy of “ignoring” a disease that’s just not that fatal is working great, and that unlike the rest of the world they’ll never have to deal with it again as they’re en route to developing herd immunity. [2]

IMO it’s becoming more and more clear that we collectively lost our minds in panic over something that’s just not that bad. The Swedish model is one to replicate.

[1] https://aatishb.com/covidtrends/

[2] https://www.google.com/amp/s/www.bloomberg.com/amp/news/arti...

Do you even check your numbers? Sweden has over 1500 deaths to date.

Your just cherry picking numbers and extrapolating to fit your conclusions. Their CFR is 10% now. 10%. That's extraordinary. You can't dismiss their current stats and then say they don't have a genpop testing regimen as a means of supporting your claims.

And even if you get "lucky" with a crude CFR of 1%, that's disastrous. If just 50% become infected, that's 50K deaths, not 500. And another 250-500K hospitalizations with long lasting effects.

(1) The books you claimed I was cooking were typos (notice they were off by one character each?). I've updated the post to reflect what I intended, so thanks for calling that out.

(2) Let's use our noodle for a sec greedo.

They haven't been attempting to control the spread of COVID, and you're telling me they only have 15,000 cases out of a population of 10,000,000 when the R0 is between 2 and 3? Do you genuinely believe that it's not a lack of proper testing? When Gangelt had a seropositive rate of 15% and Santa Clara has 2-5% previously infected?

That would make the denominator not 15,000 but 200,000-1,500,000 or 0.75% to 0.1% which is again in line with IFR estimates we've been seeing lately. Worse than the flu, certainly, but not devastating.

Frankly, that's probably a conservative denominator.

Second as I explained in a peer post, trying to establish the infection fatality rate of the disease based on its case fatality rate is like trying to figure out how risky skydiving is by setting your denominator at "whoever shows up in hospital with skydiving injuries" and the numerator at "whoever survives" -- it's peak adverse selection bias. Can you imagine the government of Sweden telling the population that 1 in 10 of them will die and them being okay with this plan? haha.

Studies are pinning the infection fatality rate at between 0.25% and 1%, and more often than not in the lower quartile of that band -- and almost all of those deaths are the old and sick who should be isolated and protected.

Further, because so many people don't have symptoms and it's so contagious, and the vaccine is at least 12-18 months away, we are all going to get it. Well 70% of us anyways. Sweden is front-loading this burden. Basically none of the numbers are directly comparable, as the minute the US re-opens it'll be playing whack-a-mole with city/state level shutdowns as China is.

Sweden will not.

Not really actually! In your age category 50-59 your case fatality rate is 1.3% [1] and given the way it affects the older end of the band more significantly than yourself in the middle of the band, it wouldn't be at all unlikely that your case fatality rate is right around 1%.

Further, we have been seeing anywhere from 30-80% of cases are asymptomatic or result in mild flulike symptoms and therefore not accounted for in case fatality rates. Indeed the Gangelt study put the overall infection fatality rate at 0.37% when Germany's case fatality rate is around 2%. It wouldn't surprise me at all if your risk was about 7X lower than the stated 1.3%, as low as 0.25%

I also wouldn't be so quick to assume car crashes are a pleasant way to go.

[1] https://www.worldometers.info/coronavirus/coronavirus-age-se...

> In your age category 50-59 your case fatality rate is 1.3%

Most of the passengers on the Diamond Princes cruise ship were 65+ years old.

Only 0.23% of the passengers died after Covid-19 outbreak.

I worry about car crashes too! :-)

That's a very reasonable reaction.

Germany's CFR is north of 3% now, and steadily climbing.

CFR is not IFR, it's not even particularly helpful or meaningful because it varies so much by demographics and hospital capacity.

> It's also worth calling out the flu has killed 2X as many people as COVID has to date this flu season.

No, it hasn't.

> A bad flu season in the US kills about 60K people each year.

You're not counting covid-19 deaths and flu deaths using the same methods so you can't compare yet. The method you're using for flu is inclusive and will tend to overcount. The method you're using for covid-19 deaths excludes some deaths and we know it's undercounting covid-19 deaths.

When we get the figures for care home and nursing home deaths included we'll see large increases in covid-19 deaths.

It is dishonest to point to a severe flu season and talk about "flu killing 60k people per year". The range is 12k to 61k. https://www.cdc.gov/flu/about/burden/index.html

> No, it hasn't.

Yes, it has. Citation needed, because I provided data and you didn't.

> You're not counting covid-19 deaths and flu deaths using the same methods so you can't compare yet. The method you're using for flu is inclusive and will tend to overcount. The method you're using for covid-19 deaths excludes some deaths and we know it's undercounting covid-19 deaths.

[citation needed]

> When we get the figures for care home and nursing home deaths included we'll see large increases in covid-19 deaths.

That's a totally dishonest way of approaching this when we know for a fact people under 50 have basically no risk from COVID and that folks who are over 70 have easily two orders of magnitude higher risk. In Italy, 99.2% of people who died were age 80.5 and had an average of 3 underlying diseases.

Age 30-39 they had 5 deaths, under 30, 0 deaths at the time this article was written [1] as compared to 852 deaths aged 80-89.

You and I both know the young aren't affected meaningfully, and for them, they won't go to the hospital with a mild cough or with no symptoms at all and so aren't counted in the denominator. You're advocating growing the numerator without growing the denominator to make your case seem stronger than it is.

> It is dishonest to point to a severe flu season and talk about "flu killing 60k people per year". The range is 12k to 61k. https://www.cdc.gov/flu/about/burden/index.html

I don't think it's crazy to compare a "bad COVID year" to a "bad flu year" do you? Our response to a bad flu year is exactly the same as our response to a good flu year. Most people don't really know there is such a thing.

[1] https://www.bloomberg.com/news/articles/2020-03-18/99-of-tho...

> Given you develop immunity after (no evidence you don't yet)

Not no evidence [0][1], it's currently my biggest fear about this virus. There's a few possibilities here:

* The tests are picking up dead virus remains

* Their tests were bad and they didn't actually recover

* Their body didn't create enough antibodies and the immunity was weak

* The virus is biphasic - symptoms stop for a while only to return later, and they never actually recovered

My hope is that what we've been seeing turns out to be among the first causes here, but we've been hearing about possible reinfections for months.

[0] https://www.reuters.com/article/us-china-health-reinfection-...

[1] https://www.npr.org/sections/coronavirus-live-updates/2020/0...

Indeed but that's not evidence of re-infection. That's evidence that re-infection should be thoroughly investigated. Rather, it's evidence of smoke not evidence of fire.

A 1% chance of death that’s evenly distributed across 80 years is dramatically less impactful to one’s expected quantity of life than a 1% chance of death that happens immediately.

Additionally, I’d note that nearly all higher income individuals take steps to reduce their auto death risks including buying safer vehicles, living closer to work, and using professional drivers when drunk or otherwise impaired.

As for your final paragraph, it is wholly unreasonable to imply that covid (which is killing thousands of Americans per day despite a massive distancing effort) is less serious than the flu (which has persisted over a much longer period of time and which was not subject to special responses).

> As for your final paragraph, it is wholly unreasonable to imply that covid (which is killing thousands of Americans per day despite a massive distancing effort) is less serious than the flu (which has persisted over a much longer period of time and which was not subject to special responses).

I didn't say or imply that. It's worse than the flu, probably 5-10X worse. Not bad, not great.

I'm suggesting that people are freaking out because there's "bodies on the street" and we're all in serious danger (we're overwhelmingly just not) when the death toll to date is about half of what we paid zero attention to whatsoever over the same time period.

My argument is that we shouldn't be nearly as freaked out about COVID as we are, and we should be more freaked out about the flu.

There's reason to take precautions, especially if you're in a risk category, but not at all to the extent of the panic we've seen to date.

In fact Sweden's plan of literally just that is going great, they've not shut anything down, and told people to stay home if they're sick (shocking, I know) and the serious-case infection growth rate has flattened out by itself. [1, 2]

[1] https://www.bloomberg.com/news/articles/2020-04-19/sweden-sa...

[2] https://aatishb.com/covidtrends/?location=Canada&location=Sw...

>your risk of dying from a COVID infection is 0.37% per the latest study from Germany

1.: Study is not complete yet and is of 500 people in the hardest hit area in Germany[1]

2.: Why do you compare deadly car accidents in the US with COVID-19 mortality rate in Germany?

3.: COVID-19 Mortality rate in the US is 5.3%[2]

[1] https://www.tagesschau.de/regional/nordrheinwestfalen/corona...

[2] https://coronavirus.jhu.edu/data/mortality

Re 1: Hardest hit would be the most helpful place to study. And I look forward to continued updates. The Stanford study, while there's certainly questions, has numbers in line with the Gangelt study. [1]

Re 2: Because much of the audience here is American, and there's not a similar study of the US population.

Re 3: No, the COVID-19 mortality rate is not 5.3%. The case fatality rate is 5.3%. Those are not the same, for an obvious reason -- currently the overwhelming majority of tests are carried out on people who go to hospital because they're sick. It's adverse selection bias.

For instance, it'd be like trying to measure the fatality rate of skydiving by measuring the odds that someone who ends up in the hospital with a skydiving injury dies. You'd think it's 99% fatal, but of course, it's not.

All you know is the numerator, not the denominator.

[1] https://www.stanforddaily.com/2020/04/17/santa-clara-county-...

That death rate is skewed. A COVID death is counted if the person had the infection and died, even if a co-morbidity would have killed them anyway. A person that has no symptoms or mild symptoms isn’t necessary known to have had the disease, which means there is a significantly lower death rate because the known infections is what’s used for the death rate, not the actual number of infections. A newly released study showed that 4% of Santa Clara County had exposure, but the “confirmed cases” is vastly lower than that. A 5% mortality rate is just false. Confirmed infections can’t be used to determine mortality because the number of actual infections is obviously going to be much higher. Mild or asymptomatic people aren’t going to the hospital to be tested so they aren’t being included in mortality calculations.

That doesn't matter much when there is no strong evidence that to have had COVID guarantees immunity[0] because being asymptomatic now doesn't mean they will be asymptomatic in the future, just South Korea has reported 116 people re-infected[0]

[0] https://www.telegraph.co.uk/news/2020/04/17/no-evidence-peop...

[1] https://www.aljazeera.com/news/2020/04/200413110301074.html

It's much more likely those in Korea received false negative test results earlier on. Let's say that was the case, though, 116 out of 2.5 million rounds to zero and is of no practical consequence to herd immunity or the potential effectiveness of a vaccine. All we need for either is 60-70% of people ending up with immunity -- so far your number is 0.0046%

All of those are great but COVID has done that in weeks.

38,000 people die every year in crashes on US roads. COVID deaths are at 39k and the first COVID-19 death was on Feb. 29. 50 days ago.

I actually dedicated much of my comment to addressing that, specifically pointing out that COVID represents a lifetime risk that's been front-loaded and so it's fair to compare COVID's infection fatality rate to your lifetime risk of dying in a car accident. I also called out explicitly that COVID has been front-loaded.

Also, it's not a fair comparison between the flu and COVID because the flu has that unfortunate feature known as horizontal gene transfer. That's why you can get re-infected by the flu each year, and why in spite of a massive vaccination campaign the flu causes 45,000,000 illnesses each year in the US alone (and 60,000 deaths) -- and 650,000 deaths worldwide.

Each year. And each upcoming year. COVID will, based on what we know so far, happen once.

You also have to take into account the uncertainty in your risk evaluation. The fatality rate is largely unknown so it's a bit deceptive to quote numbers like 0.37% as if it could be measure up to the second decimal. The German study probably came with error bars that were larger than that. For the age group 50-59, maybe it's 1.3% according to NY state (for both males and females so if the person you responds to is male, it's higher), but it could very well be 2% (or 0.5%) without it being surprising.

There is also a big assumption in considering the risk to be front-loaded. This is not the flu, but it could very well come back next year or in ten years in a different form. We just don't know yet this virus with enough confidence to do this assumption. Anyway, it is a bit strange that the ability for this virus to mutate next year would impact the risk of dying by going out tomorrow which, I think, the person your responded to was thinking about.

Finally, you're saying the flu deaths this year is double what the coronavirus has done, but again the uncertainty is high: the CDC estimation for the flu is from 24,000 to 62,000 deaths this season [1]. So it could be the double, but it could be actually lower. Let's not do things like chosing the estimate that better suits the argument without saying it's a high end of the estimate.

[1] - https://www.kcrg.com/content/news/Americas-2019-2020-flu-sea...

> This is not the flu, but it could very well come back next year or in ten years in a different form.

There has been no evidence of substantial mutations so far in COVID. There's no evidence that COVID will come back in a new and different form, any more than there's evidence I'll come back next week as a velociraptor. I mean, it could happen, but it's not something I'll plan for until I start sprouting scales and and a giant tail. Mutations are pretty random and the vast majority of mutations are harmful to the virus.

It's worth considering we live with coronaviridae all the time, something like 15% of the common cold is attributable to coronaviridae. [2] There's as much evidence of the common cold becoming Ebola as there is of me becoming a velociraptor.

The flu death range is likely more to do with how good a job we do guessing which the predominant strain will be in a given year, as the flu vaccine efficacy rate ranges from 10% to 60% each year. [1]

[1] https://www.cdc.gov/flu/vaccines-work/effectiveness-studies....

[2] https://www.webmd.com/cold-and-flu/cold-guide/common_cold_ca...

You are not taking at all into account the likelihood of suffering permanent life-long damage to lungs, hearth, kidneys and brain[0] which makes you predisposed to develop other painful (and sometimes fatal) conditions.

Also there is not enough evidence to believe COVID happens only once per person[1]

South Korea, the country with the best testing so far has reported 116 people re-infected[2], so there IS evidence about it happening.

[0] https://www.advisory.com/daily-briefing/2020/04/17/organ-dam...

[1] https://www.telegraph.co.uk/news/2020/04/17/no-evidence-peop...

[2] https://www.aljazeera.com/news/2020/04/200413110301074.html

> Also there is not enough evidence to believe COVID happens only once per person.

There's also no evidence it doesn't. What we do know is that the disease hasn't mutated significantly to date, and we see strong antibody response. [1] In time, we'll have a vaccine and it may require a number of booster shots, but what we know so far suggest reason for optimism, not pessimism.

[1] https://www.washingtonpost.com/health/the-coronavirus-isnt-m...

Reply on news.ycombinator.com