Hacker News Reader

Comment by mehrdadn | Hacker News Reader

> 1 in 200 chance of death (pretty high by the way)

The fatality rate of a car crash is also about 0.5% [1], but that doesn't prevent us from driving. That means there's a piece of the puzzle missing here, right?

Note that I'm not concluding this means we should go to gyms (or whatever). I'm just pointing out your analysis is missing an important factor, making it faulty/unconvincing.

Edit: Changed the statistic source, because it wasn't measuring what I thought it was measuring. (Which incidentally also gets at the same distinction I was trying to make in my comment.)

[1] https://en.wikipedia.org/wiki/Motor_vehicle_fatality_rate_in...

View on HN

That's over your entire lifetime, not a few weeks after getting infected. You don't have a 1% chance of dying in a car accident every time you drive.

You're right, but unfortunately it's beside my point. (I took "chances of dying in a car crash" to mean it was the fatality rate of a car crash, which was wrong.) I was trying to find the chances of dying given you have a car crash, not the chances of being in a car crash and dying. (Which I guess leads to the spoiler: this was the distinction I was trying to make. That you can't just draw a conclusion from only looking at the former, which is what that 1% figure is. You need to look at the chances of getting the disease as well.) I updated the quoted source.

> I was trying to find the chances of dying given you have a car crash, not the chances of being in a car crash and dying.

That seems a weird statistic to seek out, because your conclusion would then be "...and we don't try to avoid getting in car crashes to avoid that 0.5% risk of death". Which is wrong, we do and we should try to avoid crashes.

No, not really. "We" (as in normal people, not e.g. automakers) don't "try to avoid getting in car crashes" to nearly such a degree as with COVID. We live our everyday lives and drive anyway, and doing pretty much whatever we want, except maybe a little slower (due to speed limits), staying within lanes, and while avoiding imminent dangers that pop up on the road. With COVID we're told to stay home and just scrap everything altogether. The response is entirely different, and is not warranted by just comparing those statistics. You need to include the other pieces as well.

Maybe a better example would be all the people who pay thousands to climb Everest. The mortality rate there is something like 1.3%.

> though if you're disputing the specific numbers, you're missing my point

I'll bite ... what is your point? Approximately 30000 people die a year in car accidents a year in the US, COVID has killed 170K in six months. What's the missing piece you are trying to bring to the table?

The probability of dying given you have the disease (/car crash) is not what you need to look at. You need to include the chances of actually getting the disease (/being in a car crash) in the first place, is my point.

Ok. Almost 6 million US car accidents a year, just over 5 million US COVID cases, so car accidents are about 1/5 as lethal as COVID on a per capita basis. Currently the US yearly car crash chance is about the same as a COVID infection. Car crashes are fairly constant, COVID cases are still increasing this year.

I do wonder how the critical injury vs. lasting COVID complications will workout.

Yeah, that's what you need to compare, and that's why just looking at those statistics in isolation doesn't lead to a definite conclusion. But now that we're comparing apples to apples, and see the apples are comparable, we need to move on from them to look at other factors. And like you said, the growth rate is one such huge factor. So, really, I find the growth rate to be a much more compelling argument for treating them differently than the death rate. Though even that isn't airtight: for example, if wearing a mask would practically guarantee you didn't get the virus, and the spread of infection was merely due to people not taking such precautions in social interactions, then that could change everything. And unfortunately I don't know what the actual effectiveness is, so I can't comment on that—though personally I'm skeptical they're as airtight as we'd hope (even skeptical about N95).

All of which is to say: there's a good case to be made, and I'm not even against it, but (a) on the one hand, it's not nearly as open-and-shut as just comparing the (mismatched) statistics in the original comment, and (b) conversely, when there are more compelling arguments, it's so much better to make those arguments instead of comparing apples and oranges.

Car crashes don’t spread car crashes to others, silently and multiplicatively. Exponential risks are not the same as individual ones.

I think this is comparing the lifetime chance of various causes of death against the chance of death by COVID-19 complications after getting infected. In fact, the original report says it's too soon to know what the actual risk of COVID-19 related death is.

To be fair if we don't control this thing you have a near 100% chance if catching covid at some point in the next few months. Assuming we have an effective vaccine in 6 months you would then have a near 0% chance of catching covid again. This works out to the lifetime chance of death being the IFR. So between 1/1000 to 1/100 (depending on which study for IFR you look at).

If we don't develop an effective vaccine then given immunity seems to be temporary (as with other coronaviruses) most of us will catch this every year so a really high proportion of us will die of covid eventually (where eventually is in many years).

> To be fair if we don't control this thing you have a near 100% chance if catching covid at some point in the next few months.

By "near 100% chance" here do you mean the fraction of the population getting the disease, or the likelihood of getting the disease even if you take the appropriate precautions? Or to put it another way: imagine 1% of the population taking all the ideal precautions for the next year (and only going outside fro obvious necessities, like food/groceries/etc.). Will they also have a near 100% chance of getting it in your model? Or does 100% merely refer to # of infections / population?

The rate at which this thing grows took London from a fraction of a perentage to 5-10% infected at the same time in about 4 weeks. It could have easily hit 30-50% of people infected at the same time in another 2 weeks without the lockdown. Unless you literally shut in and careful wash and disinfect all your deliveries during the period when your delivery driver has a 1 in approx 5 chance of being an asymptomatic (but infectious carrier) you are nearly certain to catch it.

My memory is kind of fading w.r.t. what exactly happened in London, but I seem to recall their plan was pretty much explicitly "let's embrace COVID so we can get herd immunity"?

Herd immunity was England's plan initially - until they ran the numbers.

Then they looked at a new model and realised that if they achieved herd immunity, the predicted number of people who would die, this summer, was going to be massive.

They also realised that some of those deaths would be due to the healthcare system being unable to treat a lot of people, making the number who would die even larger, and avoidably so. Presumably hospitals turning critical patients away would also be socially & politically disastrous. This started the "flatten the curve" strategy.

So they rapidly did a U-turn towards progressively increasing lockdowns, and now decreasing restrictions.

Now with a bit more thinking behind us, we also realise herd immunity might not even work as predicted, because it's not clear what are the long term effects of the virus, including immunity and ability to be infectious to others after new exposure despite having had the virus before.

Like in the USA, people aren't complying with much consistency, and it looks obvious that many people, mostly younger, don't care to protect others. The emerging anecdotes and data about long term damage when surviving do not seem to bother them either.

Nonetheless, it looks like the strategy has made a large difference to the outcome compared with what early models predicted if there were no restrictions.

And now with the emerging data about long term harm (including evidence suggesting asymptomatic carriers who don't know they had it but may have sustained organ damage affecting them in future), we can be confident the strategy, by reducing the numbers of people who get it, will have made a difference to more people than previously thought.

This doesn't stop a few people from saying that the lockdown was unnecessary because there were only X deaths and the temporary new hospitals were not much used, when X is because of lockdown (and is larger than the models predicted). There are people with no brain apparently.

It seemed that way (I'm not certain it wasn't just incompetence in realising how quickly it was spreading). Then they did strict lockdown and worked hard to shield the most vulnerable. The shielding was poorly executed - like everything the UK government do.

Yeah so it's hard to see if we can use London as an example. Any other places that might be more representative?

They had 'spread the curve' type measures (stay home if sick and 'use common sense' etc) for 10 days. The infection rates carried on doubling during that time.

London is a perfect example of how quickly this thing can spread in a modern city and how many people get sick at once. The only reason the death rate wasn't way higher is that many many people self isolated for their own protection in early March.

I know less about New York but it sounds similar.

>The chances of dying in a car crash are 1%, but that doesn't prevent us from driving.

I don't think that statistic is even close to right. I know exactly zero people who have died in a car crash. Many of them are quite old.

The Australian statistics suggest that the odds of anyone in Australia dying of a car crash are 0.0000478 per person per year. This is considerably smaller than your statistic and so I am going to dispute it, at the risk of you thinking I am stupid.

I take your anecdote, and raise my own as I have had several family members die in a car crash, and zero die of Covid soo....

Your anecdote is not more valid than mine.

>>The Australian statistics

it seems Australian drive about 30% fewer miles per year on average;

in the US the risk of death in Auto Accident is 1 in 106, or .95% which I think is close enough to 1% to give the Grand Parent a little slack

Lifetime odds.

They were projecting 3 million may have died if we did nothing over a matter of months; 100 years worth of auto fatalities..

I think we can safely say at this point that those projections were widely inaccurate and have been dis proven several times now as they were based on a completely inaccurate and incomplete data set.

So, the CDC currently believes well over 200k people have died in the USA as a result of COVID-19. That's with the lock-downs and all the other measures we have taken.

What are the current projections on how many deaths would have occurred if we did nothing?

>incomplete data set

They were. The COVID deaths in the USA are about double what has been reported. Not a conspiracy, just a reality of how the deaths are coded due to testing constraints.

Reply on news.ycombinator.com