Comment by jMyles | Hacker News Reader

Comment by jMyles | Hacker News Reader

Parent Full thread

It is unlikely that this virus will kill more people than die in car crashes in a single year. Let's keep some perspective.

edit: Commenter below asked for citations, which is reasonable. :-)

This comment applies to the USA only.

There were 37,991 deaths classified under one of the WISQRS "Motor Vehicle Traffic Death" codes last year.[0]

Another 1000-3000 are reasonably counted also, for example, the 203 deaths coded as "Pedestrian injured in nontraffic accident involving other and unspecified motor vehicles." (this includes some cases of people injuring themselves with cars on private property, etc).

I think we can safely say 40,000, without even including any part of the long-term health effects of cars; this is just deaths from crashes.

We've had 71 deaths from COVID-19 so far[1]. We don't know how many total infections or transmissions have occurred; estimates vary unbelievably wildly because of so little test coverage. [2]

It's possible that our death rate is high, more likely that, at least so far, it is much, much lower than has been published in the media (the 8% of people over 65 figure, for example, uses only confirmed cases, which almost certainly represent only a tiny fraction of cases).

Based on these numbers and sources, I think it's a reasonable assertion that the number of otherwise unexpected deaths to result from COVID-19 is unlikely to be higher than the number killed by cars.

0: https://webappa.cdc.gov

1: https://www.worldometers.info/coronavirus/

2: http://perkinslab.weebly.com/uploads/2/5/6/2/25629832/perkin...

Deaths so far is an astonishingly bad metric to use for this, it's analogous to 63 grains of rice on the 6th day of the classic chessboard story.

If we don't do drastic measures, many experts are citing numbers in the high hundreds of thousands to over a million (example [1], which is well worth reading). But if we do act, the experience of China, South Korea, and Singapore teach us that we can do better.

[1] https://tincture.io/dispatch-4-from-the-front-lines-79c74fa6...

(I want to link https://www.linkedin.com/pulse/notes-from-ucsf-expert-panel-... which actually does give numbers, but that might no longer be available)

It is not the case that "many experts are citing numbers in the high hundreds of thousands to a million people". From the blog post you've cited (and which I agree is thoughtful, and which I had already read when another friend sent it my way a couple hours ago): "There are no experts. There are only good people trying their best to sift through the raft of information coming in extremely fast."

But the numbers you've cited do not represent a mainstream view in the papers currently being reviewed and published as far as I can tell.

Give this a read, "2019-novel coronavirus (2019-nCoV): estimating the case fatality rate—a word of caution": https://smw.ch/article/doi/smw.2020.20203

...and for that matter, take some time to consider generally the resources on CIDRAP; they are one of the few academic centers that seem to be actually focused on science instead of hype. http://www.cidrap.umn.edu/

Thanks for the article link and the reference to CIDRAP. I'm trying to take in a lot of information, especially views dissenting from the mainstream. These are useful resources.

> It is not the case that "many experts are citing numbers in the high hundreds of thousands to a million people".

You are objectively wrong on this. The UCSF panel on 3/10 presented an estimate with a range of 1.5 million on the high side. Dr. Osterholm said on the Joe Rogan podcast, "We conservatively estimate that this could require 48 million hospitalizations, and over 480,000 deaths over the next 3-7 months." The Imperial report warns of up to 2.2 million in the US, and even with mitigations in place a large number of fatalities. Not one of the epidemiologists I follow on Twitter has called this estimate into question, and many have praised the report.

Nobody has the real answers, and we all have to do our best interpreting the evidence that's out there, but making shit up is not helpful.

ETA: I found a working link to the 3/10 UCSF panel: https://tincture.io/an-expert-covid-19-panel-from-ucsf-664a3...

edit: OK, wow, I realize that I did indeed misinterpret Osterholm's statement on the podcast - at first listen, I thought he was discussing worldwide numbers. You are correct; his warning, based on the American Hospital Assocation, is 480k deaths in the United States.

As best I can tell, he is citing the "leaked webinar" for hospital prep from the American Hospital Association. I can't find this document or other interpretations of it.

But yeah, I was dead wrong about what he said. ---

Ahh, gotcha.

I think there's some confusion about what we're talking about.

I thought you were talking about a number of deaths in the USA (ie, compared to my pointing to the approximately 40k traffic deaths).

480k deaths, the CIDRAP estimate cited by Osterholm (which, by the way, I can't find a PDF for - do you happen to have a link), is unlikely to include more than 40k US deaths; I don't think there has ever been a pandemic where the US accounted for 9% of the worldwide deaths, has there?

For example, the 2009 H1N1 outbreak saw 284,000 deaths worldwide, with 12,000 in the USA. [0]

> The Imperial report warns of up to 2.2 million in the US, and even with mitigations in place a large number of fatalities.

We might need help interpreting the 2.2 million number - it is described as the calculation of an "unmitigated pandemic", and appears to take the worst death rate for each age group observed worldwide and combining it with the "cumulative ICU cases trigger" (the report loses me here), and then simply applying it to the USA population.

In other words, I don't think it's meant as a prediction, but a ceiling to define the parameters of the report.

But again, I'd appreciate some help interpreting.

0: http://www.cidrap.umn.edu/news-perspective/2012/06/cdc-estim...

My source for that quote was this transcript of the podcast: https://www.rev.com/blog/transcripts/joe-rogan-michael-oster...

It's 64 grains

(2 ^ (N - 1)) grains of rice on the N'th square.

That means (2 ^ N) - 1 cumulative grains of rice on the N first squares.

So 63 total grains of rice on the first 6 squares is correct.

> We've had 71 deaths from COVID-19 so far

It seems unadvisable to look at exponential curves with a linear mindset…

Consider a different way of estimating:

There are >300m people in the US.

It's easily conceivable that >50% of them could catch the virus.

CFR is likely to end up >.5% (Korea, with excellent medical care and very widespread screening, so probably one of the smaller numerators / largest denominators around, still currently has rates higher than that).

That would already give a rather high number of fatalities. And if the health care system gets overloaded, CFR is going to be quite a bit higher than assumed above.

[Disclaimer: Not medically trained or otherwise qualified]

You are probably baking in assumptions like "people who need critical hospital care but do not have COVID-19 can still receive it", which are deeply under question right now given the number of expected hospitalizations from the virus.

One of my dads friends just went through the first set of rounds of Chemo for lymphoma. They are having trouble figuring out when he can do his second set.

[citation needed]

That reallllllllllly depends on how we respond as a society in the coming weeks...

> It is unlikely that this virus will kill more people than die in car crashes in a single year. Let's keep some perspective.

I wouldn't be so sure. Italy may hit that amount this week. Italy has 3,333 deaths from car accidents per year. They currently have 2,158 deaths from covid-19, adding 368 just today.

You seem to be really bad at math. Hint: This is going to be MUCH larger than you seem to think even with extraordinary containment which is being put in place far to late.

Reply on news.ycombinator.com