How much ‘normal’ risk does Covid represent?
medium.com
medium.com
Yes. If I get all the medical care I need, then I have only an 0.15% chance of dying. (Well, more than that, because I'm male, and our numbers are twice as bad.)
But I have a 4.9% chance of being hospitalized, which generally means supplemental oxygen and an IV. And if we let this disease just burn through the population, then my state would be looking at something like 13,000 hospitalizations for 700 beds. And the numbers get much worse for older people.
Personally, I could live with an 0.15% chance of death. But a 4.9% chance of getting stuck in a field triage hospital with inadequate care, struggling to breathe? Not so much. And what about my older relatives?
You are not taking into account the asymptomatic rate, among other mistakes.
My mom had had an upper respiratory problem for about a month and they still haven’t tested her. She was just given some antibiotics a week ago; but, even if she has a bacterial infection now, couldn’t it have started as a viral one to make her sick enough to get the bacterial one this bad?
We’ll have to do statistics analysis in the coming months and years to get some real numbers on this. As of now, we’re still missing lots of data.
Hospitalization is only required if you're having problems breathing, or experiencing ancillary health problems such as pneumonia.
I'm using the Imperial College of London hospitalization numbers, from the table quoted in the article. These give a hospitalization rate that's about half that of earlier tables. So I'm guessing they probably include around 50% mostly asymptomatic cases, which is in line with heavily-tested populations like the Diamond Princess and South Korea.
Which means, in the absence of drastic curve flattening measures, a 2.5% chance of half-drowning in my own lungs in a hospital corridor full of cots. Still sounds pretty awful.
The key number we're missing is how many people catch the virus, but who defeat it quickly enough that they never show up on an RNA test. We'll need antibody surveys to figure that out. The Diamond Princess had 712 positive out of 3,711, but some people were tested fairly late. So if we assume half the ship caught it, that could cut the numbers in half again.
Actually by a factor of four if we assume that the virus managed to reach everyone it was going to reach on the cruise ship and doesn't spread better on land.
And if only those 712 caught it, that reduces your number by a factor of five. So we are already down from 4.9% to between 0.5% and 0.625-1.25%.
Looking at the data from https://www.worldometers.info/coronavirus/
It seems logical to state, at least within the US, the accuracy of deaths due to COVID-19 are more accurate than total cases of COVID-19. There are a lot of asymptomatic people out there. Even if someone has it with mild symptoms they are told to stay home and only reach the hospital if symptoms are severe. Deaths are more accurate as I assume a test can be performed posthumously for COVID-19.
So, even with this exaggerated ratio, the US has 5 deaths/1M pop. One can reasonably argue this might've been worse if no quarantine measures were in place. However if the true rate is less than 1%, I think the economic effects will be more impactful.
The above link also shows ~27k deaths worldwide due to COVID-19 at the moment. Italy had ~25k deaths due to the flu alone in 2016/2017.
You have forgotten about serious irreversible damage to your lungs which will probably shorten your life by a few years.
The data in the article tell us that most of those 446 deaths in which C19 played a part would have happened in the next few years anyway.
So we would expect fewer people to die in Bergamo next year, because they died this year instead. And then fewer than normal expected the year after, and so on, although it gets increasingly noisy with time.
We’re measuring the impact of COVID in deaths, by and large. A more useful unit for some purposes is quality-adjusted-life-years (QALYs).
For people who die, the QALY impact of C19 would have been how many years of full health the person had left, if they had not died after catching the virus.
But similarly, for people who don’t die shortly after contracting it, C19 can still have a QALY impact, and we’re not reporting or even measuring that much. Patients often have reduced lunch function even after “recovering”; there will be patients in whom C19 triggers post viral fatigue, which can last years; in the worst cases patients will be stuck with ME/CFS or similar for the rest of their lives.
Very little about the philosophy, economics, politics or science of dealing with C19 is simple. Although clearly some things are: if I don’t go near people with the virus, I won’t catch it; if we flatten the curve, fewer patients will die; the old and the infirm are more at risk. But those few simple parts don’t necessarily make the whole simple.
So overall deaths per capita will probably still be much higher than before for the next 5 to 10 years. Maybe longer.
So it doesn't disprove the hypothesis that the average person who gets it has a 1 % chance of dying.
If the virus causes one year's worth of extra deaths, these will not be spread out over the whole year, so seeing 350 of them in a month of the epidemic would also not seem out of place, it corresponds to an epidemic whose peak is a few months long. So this data doesn't seem inconsistent with the linked article's estimates.
Of course if you are running a morgue, then 448 instead of 98 is the relevant figure, and is likely to be very difficult to handle. And if you are running a hospital, even worse.
> So, roughly speaking, we might say that getting COVID-19 is like packing a year’s worth of risk into a week or two.
Sure, but it also means these deaths are piled up on top of everything else. It's not like people will stop dying of other causes.
> many people who die of COVID would have died anyway within a short period
This strikes me as completely unfounded. Diabetes or an age > 70 aren't imminent death sentences in a developed country.
> if COVID deaths can be kept in the order of say 20,000 by stringent suppression measures, as is now being suggested, there may end up being a minimal impact on overall mortality
In a sense this is a tautology. Of course, if we keep deaths to a low number there won't be an impact on mortality rates.
You're misunderstanding: the people who die of COVID-19 will stop dying of other causes, because they've died of COVID-19. There will be a decrease in other-cause mortality...
>In a sense this is a tautology. Of course, if we keep deaths to a low number there won't be an impact on mortality rates.
...thus what it means is that if the COVID-19 deaths are kept to people who are ill enough that they would have died within the year anyway, those deaths won't be extra mortality across the year. If on the other hand COVID-19 spreads a lot and kills people who wouldn't have otherwise died, there will be extra mortality.
If you compare deaths directly, you get really horrific numbers.
For example: Covid will kill significantly more people in the US than WWII and the deaths will occur within roughly a year or year and a half. (WWII lasted 4 years for the US but the popuation was also smaller). If that sounds horrible and unacceptable, consider that tobacco related deaths (including second hand smoke) in the US kill more than WWII every year. https://www.cdc.gov/tobacco/data_statistics/fact_sheets/heal...
Because the deaths will come in short time span and hospitals will be out of capacity the situation will be warlike in countries that fail mitigation. Hospitals running out of capacity kills more of people than the disease. Isolation failures due to hospitals not having testing equipment and protective gear will add unnecessary deaths too. The risk from the disease itself is comparable to lifestyle risks and many other risks.
Where I am, I don’t see less car traffic. We are moderate in terms of US states for cases. We are told to only leave the house for urgent matters. But the chic-fil-a drive through line is 45 cars deep.
In the US automobiles kill roughly the same amount of people annually as year in the Vietnam war. Something like 30k+ annually. You can actually subtract some car deaths from covid deaths because there is less traffic.
9/11 or terrorism in general so low risks that it's quite clear that responding for terrorism with conventional war and trillions of dollars is insane.
There have been only a few hundred thousand cases of COVID19 so far, and thousands of deaths. The fatality rate is many magnitudes greater.
Can anyone look into these criticisms and see if they hold weight?
One thing this plot does very well is put into perspective the extra risk to the elderly.
https://www.theguardian.com/world/2020/mar/26/men-are-much-m...
I’m not saying the answers should influence our policy decisions necessarily, but it’s still an interesting question.
[1] https://ec.europa.eu/commission/presscorner/detail/en/MEMO_1...
It gives an idea of how complicated it can be to work out the additional mortality caused by a disease.
https://assets.publishing.service.gov.uk/government/uploads/...
Any idea what this "clear" cause is?
[1] https://en.wikipedia.org/wiki/List_of_countries_by_traffic-r...
[2] https://en.wikipedia.org/wiki/List_of_countries_by_suicide_r...
just my 2ct
11,658 cases as of writing. 578 deaths, 135 recoveries. 163 in serious condition.
1.4% critical, 4.7% deaths, that's not too bad.
But what worries me is that statistic with 578 deaths and 135 recoveries. This is probably because there's a lot of unreported cases, maybe 10,000 people feel sick, stay at home, don't want to tell anyone, and a portion of those die because it's too late, or have an asthma attack classified as Corona. But it looks like a 80% mortality rate from this really pessimistic angle. What am I missing here?
#deaths/(#deaths + #recoveries) will overestimate CFR, and #deaths/#cases will underestimate CFR. The final result will be somewhere in the middle, around 1-1.5%, unless the hospital system gets overloaded in the UK.
They aren't even testing all people with symptoms yet here. Only if they are severe or the patient is high risk.
The UK is only testing people admitted to hospital for suspected covid-19, and it's not testing all of them.
PRC is new model of what we should aspire to.. because they stopped the pandemic(after covering it up and letting it spread first)
When your parents put you in timeout did you equate it with being imprisoned?
That attitude that the government is like a parent and the citizens its child to be disciplined, as opposed to a humble servant to the citizens, is exactly the kind of totalitarianism the poster was referring to.
My point was that equating quarantines with totalitarianism is the kind of superficial and querulous comparison that a child might make about timeouts.
My point is that it's generally accepted that children have less rights than adults. E.g. it's accepted for parents to lock their kid in a room for hours for swearing, but it's not accepted for the government to do the same thing. The fact that it's socially accepted to treat a child a certain way doesn't mean it's socially accepted for the government to treat adults the same way.
Similarly for the argument about not having enough hospital beds due to other people getting infected: the classical idea of rights was as "negative rights", not "positive rights". A negative right means you have the right that somebody else will not do something to you, whereas a positive right means that you have the right for somebody else to do something to you (or, somebody else has the obligation to do something for you). Positive rights necessarily impinge on negative rights. Since requiring hospital care is a positive right, not a negative right, approaching it from a system of negative rights one wouldn't have any right to restrict other peoples behaviour under the justification that it would improve your access to hospital care.
Similarly, you have the right of free assembly, but not when it puts others' lives at risk.
And regardless of the legal/moral distinction, I would suggest that these limitations are both legal as well as moral.
That's an interesting example, because that quote was actually originally used in the context of jailing somebody for opposing the WW1 draft: https://en.wikipedia.org/wiki/Shouting_fire_in_a_crowded_the.... And the actual ruling was partially overturned in https://en.wikipedia.org/wiki/Brandenburg_v._Ohio. There's an in-depth analysis here: https://digitalcommons.chapman.edu/cgi/viewcontent.cgi?artic... . In short, it's not clear that you don't in fact have the right to shout fire in a crowded theatre.
Personally I don't support quarantine because I believe a negative-rights-based system produces the best outcomes in the long term. Yes it might do worse in a pandemic like this, but it also severly limits the possibility of other worst-case outcomes like the hundred million people killed by Mao and Stalin, the Cambodian genocide, Hitler's atrocities. If the countries involved (and by extension their governments) had been absolutely committed to the protection of such negative rights (and had institutions that made it extremely hard to violate these rights, a strong court system and constitution), they could never have justified the atrocities they committed.
And government will never be your humble servants no matter how much literature you will produce on the subject. They have tanks, you do not. Shepard is never a servant of the sheep. Even though he feeds them and keeps them safe and healthy.
The best government is the one that harms the citizens as little as possible in the process of taking value from them to give it to the people that sit on all the weaponry, so they don't take whatever they want with that weaponry. Everything else is a sideshow aimed at improving the stability of this setup.
It's not as simple as stupid. Humans have different values and risk preferences, and all values are ultimately arbitrary so there's no objective way to say some values are better than others (https://en.wikipedia.org/wiki/Regress_argument). In that case people have highly conflicting values in terms of weighing up loss of life against destruction of peoples' standards of living. People also have conflicting estimates about how much economic damage the shutdowns will cause, so even if they did have the same values they wouldn't necessarily agree.
>And government will never be your humble servants no matter how much literature you will produce on the subject. They have tanks, you do not. Shepard is never a servant of the sheep. Even though he feeds them and keeps them safe and healthy.
This isn't exactly an argument that governments aren't behaving in an authoritarian way..
Most relevant knowledge at the moment comes from epidemiology.
The most important value we have to take into account at the moment is that suffocating to death is very bad and we want to inflict this upon least amount of people possible.
Money is just a way of keeping tabs of to how much of the output of our civilisation every entity is entitled to.
We did it sort of dilligently in hopes of encouraging people to increase the output of civilisation. It kinda works but also led us to dog tailors, Jeff Bezos private wealth and cruise ship industry and various other crap so there's definitely room for improvement.
Lots of people hope that economy won't crash but will be knocked out of this local optimum it sits stuck in for at least few decades.
> This isn't exactly an argument that governments aren't behaving in an authoritarian way..
This is an argument that they always behave in authoritarian way. The only thing you can hope for is that they bother you the least amount possible while doing their thing which is taking part of your output.
You had rights because it was beneficial for the government to give you rights. Epidemic changed that. When it subsides if it will still be beneficial for the government to give you rights back, you'll get them.
If government fails, army will step in to take what they want from you more directly.
Putting that aside, let's assume that the epidemiology of this disease is settled. What would you do differently policy-wise to address this, given that you seem disinclined to favor suppression methods?
Let's even assume incorrectly that all COVID deaths are on top of the 600k normal deaths. So in 2020 85% more people die of natural causes than normally. Almost exclusively very old and already ill people. It's just not a big deal and it's crazy that people think it is.
Now let's say that the death rate is 0.2% as the Oxford institute estimates, 20% of COVID casualties would have died anyway, and that isolating the at-risk people further reduces the casualties by half. We end up with 40,000 extra deaths, 7% more deaths than normally. A non-event.
If you think that the death toll of the lockdowns will be lower than that, then I hope you are right. I expect not.
[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...
The Oxford CEBM estimate is an extreme outlier, and I have not seen a pre-print of any rigorous work that was used to produce those estimates. That, for lack of a better word, blog post you linked to is full of spurious analyses like the following:
>Therefore, to estimate the IFR, we used the estimate from Germany’s current data 22nd March (93 deaths 23129) cases); CFR 0.51%
So they essentially just took the current cases and produced a naive case fatality rate, despite the fact that many cases are still ongoing? By the way the number of deaths is 281 now and Germany's nCFR is climbing quickly, yet they have not revised their estimate higher. Why is that?
I'll also note that they conveniently exclude South Korea from their analysis. Why are they only choosing countries that favorably support their "low IFR" hypothesis?
>The results of screenings have suggested 0.5% are infected; the true figure is likely higher due to asymptomatic and as a result of many not seeking testing: estimates suggest the real number infected is 1%. Iceland is currently reporting two deaths in 737 patients, CFR. 0.27%; if 1% of the population (364,000) is infected then the corresponding IFR would be 0.05%.
They neglect to include in their analysis that the 1% infected ratio for Iceland came from a voluntary screening study, meaning that it was not a random population sample but most likely enriched for infected cases. They then naively back this out to a ridiculous IFR of 0.05%. Honestly, are these people even actual researchers?
>Let's even assume incorrectly that all COVID deaths are on top of the 600k normal deaths. So in 2020 85% more people die of natural causes than normally. Almost exclusively very old and already ill people. It's just not a big deal and it's crazy that people think it is.
This ghoulish statement is all that I really need to know about you.
The same question could go for you, why do you only listen to specialists that are contrarian outliers?
People are whining about volunteer isolation after a couple days. I've been living it for roughly 6 months. And my self quarantine reduces my individual risk while a disease spreads fast. This is just the beginning for normally healthy people.
> So, roughly speaking, we might say that getting COVID-19 is like packing a year’s worth of risk into a week or two. Which is why it’s important to spread out the infections to avoid the NHS being overwhelmed.
These numbers, afaik, are just wild guess-estimates. The real numbers are still not known because we are yet to have a breakdown from an country that went with herd-immunity and see how that did pan-out.
But the real question is: Is the mortality rate, overboard, increase and if so, is it increasing significantly.
Italy is recording ~700 deaths per day. But ~2.000 deaths per day happens on a regular basis. If we were to pack a year of deaths into a two weeks slot, then we should see ~40.000 deaths every 24 hours. We are clearly very far from there.
It's surprising that no country is doing that but what we should do is test people randomly. That will estimate the % of the population that is currently infected with Covid-19 and determine a more accurate fatality rate.
It's very possible that the Covid is a nothing-burger really and the recorded deaths would have died naturally anyway.
In the later phases, you want to know who has been infected, yet the commonly available test in fact measures who is currently infected (ie. It can't detect those who are still liable to get infected, which is what you need for good modelling).
Antibody tests which are being prototyped by many groups should fill this need.
Can't you see how these factors mean that testing 100 people would not lead to a proper representative sample?
I don't think 100 people would be enough to properly and statistically test for a common cold, much less for a disease like this.
Where did I ever write this? I wrote it would give us an upper and lower bound.