Hacker News Reader

Comment by strangeloops85 | Hacker News Reader

Some more information about the study here: https://pressroom.usc.edu/what-a-usc-la-county-antibody-stud...

Population is supposed to be representative of LA county. Tests are from Premier Biotech with some preliminary data on false negative/ false positive rates shown. I suspect assumptions related to the test itself, rather than population, may be a bigger factor for any inaccuracies. There's also the question of whether antibody response implies immunity for SARS-COV2

Another interesting aspect of the result was 6% of men showed antibodies but only 2% of women tested.

Also: --2.4% of people between the ages of 18 and 34 had antibodies to the coronavirus --5.6% of people between 35 and 54 had antibodies --4.3% of between 55 and older had antibodies

Edit: The key assumption that really, really needs to be triply checked here is the specificity of the test. They're assuming 99.5%, similar to the Stanford study I believe. If it's closer to 98% then prevalence estimate confidence intervals should include 0%, I believe. Also should be noted that the same PI on the Stanford study, Jay Bhattacharya, is also involved in this. Same criticisms likely apply here - more caution needed!

Second edit: On further reflection, I think these studies could do serious damage based on how the results are being announced by press-release - in this case, afaik there's no pre-print. In our current environment this isn't just science. Policy is being enacted and people may die because of poorly framing results. Everyone involved needs to take a step back and think through this carefully. Also - did anyone figure out why a hedge fund guy was an author on the Stanford pre-print? And why that same guy then published a Wall Street Journal op-ed about the study without identifying himself as a co-author? Shady stuff.

View on HN

I’m very skeptical. Why were they so few hospitalizations in LA vs NYC if such a large % in LA contracted Covid? Is this suggesting that a much, much higher % in NYC have it (and the number of those with any serious symptom and thus need for hospitalization was higher)?

Yes, that is the implication. And if this is the case, that is very good news.

Yes. The consensus I've seen forming is that at least 15% of NYC has had it by now - a variety of questionably random samples have suggested numbers near that.

NY has 17x more deaths than California. That would mean if 3% of CA was infected (yes this is LA only but for ease of comparison) it would be 51%+ of NY was infected. Far more than just 15%.

CFR is not a great way of comparing across regions, it could be due to any number of factors. Demographics (i.e. age), co-moribidities and the amount of hospital space available.

In fact the per-country spread of CFR is huge, from a low of 0.07% in Singapore to a high of 14.93% in Belgium, a whole 3 orders of magnitude. The H1N1 influenza CFR at a similar point in the outbreak was between 0.1% and 5.1% -- but the final IFR ended up at 0.02% [1].

[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

The comparison is overweighting LA county, which has 25% of California's population but 60% of its deaths. (I don't know what the math ends up being off the top of my head.)

LA is one of the densest parts of CA though, outside of SF/Bay Area. So if LA is 3%, you could assume the state as a whole is maybe 1-2%. Then 17x that becomes 17-34%, which seems more reasonable.

Consensus among whom?

https://www.nejm.org/doi/full/10.1056/NEJMc2009316

15.6% of pregnant woman at a given hospital were PCR confirmed over 2 weeks. I haven't seen strong arguments this wouldn't be broadly characteristic of the overall population. Worse (or better depending on how you look at it), there's significant false negatives with PCR testing.

It's credible NYC could be at 30% by April 4 and given that confirmed cases doubled since then, could be at 50%+ by now.

You can't extrapolate from pregnant women who have been visiting medical facilities to the general population.

How off do you think this can be? Women go once a week to a medical facility (and that's to an ob-gen, not the emergency room) before birth, but also don't go out in general. You could just as well make the case this is lower than the general population.

It's for sure true that a huge number of people in NYC got it, way surpassing the confirmed cases count. The state is currently doing a serology study here and results should be published next week.

Sure I believe NYC has way higher prevalence of cases but difference in deaths would imply majority of population was infected and they’d essentially be at herd immunity now. It’s not implausible but find it hard to believe.

We'll know soon enough. The blood serology randomized trial is underway. I suspect some neighborhoods might be at 50%, but the city overall won't be.

Maybe they'll finally end this ridiculous lockdown when that one is released

I think NYC almost certainly has much large prevalence than LA given hospitalization data alone.

One factor could be the nature of transmission. Regions where transmission occurs with high viral loads entering at the point of infection (for example, hospital-acquired infections) could see more severe disease progression. You would therefore expect a high hospitalization rate relative to prevalence. It's possible in other regions, including CA, hospitalization rate as a function of prevalence is lower because acquisition is largely at lower viral loads resulting in less rapid and severe disease progression. This is, of course, a hypothesis and very likely not correct. Time will tell.

Other important factors include age distribution of infection, as well as weather conditions and prevalence of pre-existing conditions.

Yes agree NYC will have much higher prevalence but it has 17x the deaths - if fatality rate is similar then it would be like 50%+ of New Yorkers have contracted Covid. Just a little hard to believe.

5.6m people ride the New York subway everyday.

There’s a decreasing number of positive test cases each day in NYC too though. At highest point was 50% and now under 30% or so. I’m not sure if those tested get the antibody test as well but it implies a larger percentage of symptomatic people are testing negative in nyc.

Almost no one gets an antibody test yet, unless it's some special controlled study.

Actually, no they don't. There are 5.6m rides per day. The average trip contains at least two rides: there, and back again. Many trips contain substantially more than two legs, and many people take more than one trip per day.

I can't tell you how many people ride the subway every day, but it's certainly less than half of the number of rides per day.

Number of rides is actually more relevant in this context than number of distinct riders. Each trip involves a new set of people next to you in the train car and a new chance of getting infected.

The number of rides is useless in estimating the percentage of the population with high risk for exposure if you don't also have the proportion of the population exposed to that risk.

Closer to 2 million: https://nyc.streetsblog.org/2020/03/23/the-subway-ridership-...

What makes it hard to believe, exactly?

One way to explain the hospital surge in NYC is that a small percentage of people have had a highly fatal disease. Another plausible explanation is that a much larger number of people have had a less-fatal disease.

To date, we've had very poor estimates for fatality rates and prevalence. It wouldn't be shocking at all to find out that we're off by a factor of 10-, 20-, or even 50-fold.

Typically a sufficiently infectious disease tops out around 50-80% (don't have a source, just what people, like Angela Merkel have been saying). So this implies NYC is already off the high end of that. Hard to believe.

NYC has had ~141,000 infections confirmed. If the high end of this study held for NYC (50x), then the true infection count would be around 7M.

That isn't impossible at all, given that the population of NYC is around 8.8M people. It would mean that the city is approaching herd immunity.

(I don't personally think the ratio will be as high in NYC as in LA, but that's just a hunch based on the higher testing rate.)

That's basically the top end of the estimate of what is theoretically possible without a lockdown and infinite time. I just don't buy that it is already maxed out like that.

It's the top end estimate from the study. Even if you take the low-end estimate (~30x), there would be 4.2M cases, which is still ~50% of the New York population.

Basically, you're just saying that you don't want to believe what they're saying, therefore it's hard to believe. But there's nothing illogical or impossible about it in the slightest.

Even 50% is hard to believe. Illogical or impossible, no. Improbable, I think so.

You keep making assumption. You have to concluded the lock-down did nothing, has done nothing, and hasn't prevented any infections. You can make assumption because the numbers don't align to what you assumed they'd be.

If what you're saying is true, then the NYC death rate will drop to zero in three weeks.

I am willing to bet the farm that it won't.

Yeah, likewise.

That seems weird since all data points to that being whats happening.

What data?

How is that hard to believe that incredibly obvious with the dropping hospitalization rates

I also suspect and absolutely think a large portion of the reason there is a decrease in hospitalizations is due to mitigation efforts decreasing the initial viral load. That is absolutely borne out of all evidence we know thus far for this virus and is also in line for what we know to be true about other viruses.

Excellent point!

In part because all the headline numbers we've seen are CFR numbers which suffer from huge adverse selection bias -- and aren't directly comparable between regions for a lot of reasons.

As I've mentioned before, using CFR to determine fatality rate of the disease is like trying to determine the fatality rate of skydiving by measuring the ratio of people who go to hospital with skydiving injuries over those who recover. Obviously you'd think skydiving kills basically everyone who does it, but it just doesn't.

It's starting to become clear that the infection fatality rate of COVID is in the lower quartile of the range 0.25-1%.

And yeah, a huge portion of New York likely has it. It's super contagious.

Define "huge portion" because will still have to square this with the negative test data from NYC.

Yeah, trying to do the basic math on the implications of these figures...[Not claiming they're true, that's a different question]

The article says 221,000 infected at a 2.8% rate, in "early April". So they're assuming a 7,892,857 population. Currently, we have 601 Deaths total in LA[1], which we can assume are a result of early april event on April 20. So a 0.002% "fatality rate" relative to total exposures, "Woohoo" (NOTE, this is an optimistic figure implied by this study, the standard 2% [edited] previously claim but still not equivalent to the flu, which has 0.2% [edited] relative people who get sick). Now, with weekly doubling: 221,0008 = 1,768,000 infection now = 22.4% - > 4808 deaths. However, past 10%, we can ask whether we start reaching saturation/diminishing returns. But that is a key question.

-- Let's look at NYC. There we have 14,500 Death. Using the article's implied figure, 14,500/ 0.002 = 7,250,000 infected on April 1, out of 8,398,748. So this implies an infection rate of 7,250,000/8,398,748 = 86.32239%, which, in fact, seems very implausible and which implies; a higher death rate among "vulnerable people". Maybe we're about to hit full saturation with NYC and see an immediate decline.

Now, the other bad* scenario is the possibility that a lot of sickness and death NOW is being driven by reinfection, that exposure isn't giving much immunity at all and instead of the unexposed dying, what we're seeing is the re-exposed dying and with this dynamic, the death rate settle into a steady and unacceptably high level. Note that Italy, which has seen a high death rate, hasn't a strong drop-off at all, just gradual plateauing. IE, guess what, one possibility is "herd immunity" just doesn't exist for this virus.

Anyway, time will tell, literally a week will tell if NYC has enough fatalities to make this 0.002 exposure-to-death rate numerically impossible.

[1] covid-19.live

Edit: corrected conversion ratio to percentages.

I think that number should be 0.2% (2 / 1000) and not 0.002% (2 / 100000).

Yeah, 2% is the standard and more plausible in many ways than 0.002% [edit, WRONG ,missed a decimal point, should be 2% & .2%]. My comments above are about what this study implies, this something new.

Italy has had 24114 deaths. If take those as being in Northern Italy with a 27,000,000 population, that's 0.8931 % of the entire population. Even adjusting for age, that can't squared with 0.002% death rate [wrong wrong, it's 0.08931%, which is possible but with a huge infection rate, which maybe].

I think you've misunderstood the comment by zucker52: your post is using 0.002 and 0.002% interchangably. It's quite hard to figure out what you mean when the numbers are off by two factors of magnitude in places.

For example:

> So a 0.002% "fatality rate" relative to total exposures,

You appear to be deriving this from the computation 601/221000. That gives .27%, not 0.002%.

> Italy has had 24114 deaths. If take those as being in Northern Italy with a 27,000,000 population, that's 0.8931% of the entire population

No it's not. 1% of 27M is 270k, not 27k.

Damn, I am being sloppy,

Yeah, 0.08931% of the population of Italy experienced fatality.

If we assume .2% fatality rate, that implies 50% of the population was infected (which might make sense with an extremely contagious virus but that's more than even the cruise ships).

The issue is you're not adjusting for the exponential fatality risk for old folks, and that those in Lombardy are the oldest in Europe. It's hundreds of times worse for old folks than for young folks.

The US doesn't have a hundred times less old people than Italy. The whole "don't worry about old people dying" attitude of those eager to "reopen the economy" is kind of despicable.

Florida has 20% of it's population over 65 and it's 21 million people might similarly be rather hard hit.

> The US doesn't have a hundred times less old people than Italy. The whole "don't worry about old people dying" attitude of those eager to "reopen the economy" is kind of despicable.

It's totally despicable, and it's not my position.

I propose we do what the UK proposed, and what Sweden is doing: keep all vulnerable folks inside, provide them with services such as groceries and any healthcare they may need on-site, and release the low risk to build a protective barrier of herd immunity around them. That should have been evident when I said "old people were hundreds of times worse off."

That's totally orthogonal to my point, which is that you can't project the mortality rate of a population that's overwhelmingly disproportionately affected, scale it linearly onto the world population and assume that'll be at all representative of the reality on the ground.

And yes, Florida may be affected very badly, and the plan may need to be adjusted accordingly.

"Now, the other bad* scenario is the possibility that a lot "of sickness and death NOW is being driven by reinfection, that exposure isn't giving much immunity at all and instead of the unexposed dying"

This would explain: NYC vs LA vs Italy. Why isn't this theory brought up more?

"Wang, whose full name has not been disclosed for privacy reasons, is one of more than 100 reported cases of Chinese patients who have been released from hospitals as survivors of the new coronavirus — only to test positive for it a second time in the bewildering math of this mysterious illness."

https://www.latimes.com/world-nation/story/2020-03-13/china-...

> Why isn't this theory brought up more?

Because if reinfection so close to recovery were possible, it would make this virus really unique, and it doesn't really fit the available data very well. It makes much more sense that the virus takes a long time to clear after symptoms abate, and the tests have a really lousy false negative rate.

The common cold is infamous for mutating at a rate where you can catch the virus again two months after infection. And moreover, the common cold is also a coronavirus.

Moreover, if a lot of Covid infections are in the area and the virus is mutating quickly, a person might be exposed to a different strain soon after recovery and wind up infected, if the SAR-COV-19 is similar to the common cold. This would be hard to notice, since it would tend to happen as the virus was spreading rapidly anyway. It would become a problem only if people expected the virus to abate once everyone had been exposed.

See: https://www.technologynetworks.com/immunology/news/why-dont-...

No. The "common cold" is a collection of 200-something distinct viruses with the same/similar symptoms, only like 10-20% of which are coronaviruses. Most of them are rhinoviruses.

There’s no evidence yet that COVID is mutating in a way that would cause reinfection. There haven’t been any documented confirmed cases, the speculative cases are much more likely false negative test results followed by actual positive test results.

If what you were saying were true then we would be seeing a huge number of symptomatic re-infections and deaths in China and Italy. As far as I know the ‘reinfections’ have been asymptomatic. That suggests that the virus takes a long time to clear, not that it’s mutating.

The main reason should be that speculating on both this and the exposed-and-maybe-immune isn't that relevant for the more responsible approaches to managing the situation - isolation, quarantine, safety equipment, testing and contact tracing.

This kind of speculation is relevant to the more irresponsible approaches; "developing herd immunity" and "immunity certificates" but it mostly should be a reminder to "don't do that because you really don't know what you're getting into".

It’s not clear this is the responsible solution as there’s no evidence the entire process wouldn’t start again the minute restrictions are lifted due to the sheer infectiousness of the disease and the large number of people with no symptoms or limited symptoms.

> Note that Italy, which has seen a high death rate, hasn't a strong drop-off at all, just gradual plateauing.

Regional differences? Lombardy's death rate is less than half of peak

What’s the source on reinfection? That sounds kind of extraordinary.

It seems weird to think the body can clear an infection but have no ability to fight it later.

There is none, it is just speculation/hysteria like so much else about this virus.

There's been only a very few cases globally where a person who previously tested negative later tested positive, for which a test giving a false negative is a much more likely explanation - especially as these cases were early on when tests were more unreliable. Otherwise - everybody else who tests negative does so consistently.

The virus is very similar to the original SARS (genome about 80% the same which is huge). It is a relatively stable by virus standards not mutating massively. In fact, if you had SARS your are likely immune from covid19: https://marlin-prod.literatumonline.com/pb-assets/journals/r...

So the general scientific consensus I have seen is that there's no evidence to suggest covid19 is some sort of supervirus that will reinfect people and defeat your immune system.

There's also been some animal testing on reinfection : https://www.biorxiv.org/content/10.1101/2020.03.13.990226v1

Yes, reinfection is speculation and limits the virus to infect the population is speculation. Beyond the situation of the virus growing exponentially, there are few certainties.

I add this point because most of speculative arguments are used to claim allowing the virus infect a large portion of population won't cause much harm.

There’s clearly a lot of false negatives produced by the PCR testing we’re doing nationwide.

According to the New York data, the fatality rate is about 0.5% for tested persons without underlying conditions. Considering the gross selection bias it likely means actual death rates in the healthy population close to that of the flu. It is this data that shows the lockdowns were a complete illegal overreaction. Locking down nursing homes would have fixed most of it.

https://www1.nyc.gov/assets/doh/downloads/pdf/imm/covid-19-d...

Illegal, no, but a massive overreaction. This is why the Swedish model (of telling people to take reasonable steps like not getting too close, and staying home when sick, and away from old people) is proving so successful. They've even started to see a flattening of the curve alongside the rest of the world without lockdowns. [1, 2]

[1] https://fee.org/articles/sweden-s-top-epidemiologist-covid-1...

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

Quarantining the sick is legal but otherwise the freedom of assembly clause in the 1st amendment does not say "except when there is a pandemic or other fear". If you don't believe that should be the law fine, but since it is law and is incorporated to the states via the 14th amendment it qualifies as illegal.

Not quite; you'd have to drag this through court to get any sort of legal determination on it.

And I think they would just say it was "reasonable" given the information known at the time.

Can you please provide links to judicial opinions or rulings that validate your interpretation of the law?

If you read the federalist papers you'll note that the intention was literal, not to be reinterpreted for the times. That also makes the thousands of gun laws illegal. If you start reinterpreting the constitution that is how you go full banana republic, and everyone knows you never go full banana republic. You are supposed to change the law not reinterpret.

It's pretty funny to watch people try to justify "literal" readings of the text, since they often focus on interpreting just one piece of text and ignore the effects of the analysis on the rest of the text.

The First Amendment very clearly binds only Congress in its literal reading: it begins "Congress shall make no law..." And there is no literal text in any subsequent amendment that might cause incorporation to the states--the Fourteenth Amendment only literally incorporates the Fifth Amendment, and that by literal repetition of the text.

The fact that legal realism--which says that the law is whatever judges say it is, even when what the judges say is patently ridiculous when compared to the actual words of the Constitution or statute--is the mainstream viewpoint in today's legal environment does not make it right.

I would argue the second amendment doesn't explicitly say you can't own a nuclear warhead for personal amusement, but I would argue nobody's going to challenge the validity of such impositions.

Probably because they know it's futile, not because it shouldn't be covered under the Amendment.

It almost certainly shouldn't be covered. It states "being necessary to the security of a free State". What part of a nuclear warhead wielded by a militia contributes to the security of a free State? Nuclear weapons are rather unique in that they're essentially useless in a civil war. What are you going to do? Nuke the very land that you're fighting over?

Biological and nuclear weapons should not fall in the scope of the 2nd amendment. They're useless for guaranteeing your liberties against a tyrannical government, and the consequences for storing them improperly are severe.

> What part of a nuclear warhead wielded by a militia contributes to the security of a free State?

Depends on who's attacking and what might deter them.

> They're useless for guaranteeing your liberties against a tyrannical government

Which is only part of "the security of a free State", not all of it. The militia was also for repelling foreign invasions.

Given our current modern world, I would agree that weapons of mass destruction (nuclear, biological, and chemical) shouldn't be available to anyone who wants them; but the correct legal way to make that happen in the U.S. would be to amend the Constitution to explicitly add that exception to the Second Amendment. As the Second Amendment is currently written, it does not admit of any exceptions. As I noted in another comment upthread, privately owned warships were significant at the time the Second Amendment was passed, and the framers of the Amendment did not exempt them from the category of "arms".

"Arms" in the context used by the second amendment is AFAIK generally understood to mean "small arms" and by implication would exclude such weapons by matter of definition.

> "Arms" in the context used by the second amendment is AFAIK generally understood to mean "small arms"

No, it isn't. In the time period in which the Second Amendment was passed, privately owned warships of similar capability to the ones in national navies were common enough to play a pivotal role in two wars (the American Revolution and the War of 1812).

I'm aware of privately owned warships by privateers. It is true and a good counterargument.

However, the colloquial use at the time of the phrase "arms" almost certainly referred to small arms, but would have included knives and swords (State v. Kessler 1980[1]) which are paradoxically much more heavily regulated than firearms in most states. See also "To Bear Arms"[2], by which the modifier "to bear" suggests arms that could be carried.

Perhaps one counter argument to this point was the desire by the framers to include a conscientious objector clause obviating military service for people who had specific objections, which may support the view that "arms" was anything in use by the military of the day, but the clause was stricken from the constitution for fear that the second amendment may be interpreted by later generations as a right that could not be enjoyed by the people and was intended to encompass military service.

There is also the argument that the framers opposed the idea of a standing army, which would support your viewpoint, and there is an excellent essay[3] that makes a fair and reasonable argument to this end.

My personal opinion is that this isn't clear, but it is obvious that--at a minimum--the interpretation of "to bear arms" would suggest that the 1986 ban is unconstitutional, as is the 1934 NFA tax act that required registration. I'm not convinced "to bear arms" includes other weapons that could not be "beared," such as warships, but perhaps this will clarify my thought process for you.

[1] https://law.justia.com/cases/oregon/supreme-court/1980/289-o...

[2] https://guncite.com/gc2ndmea.html

[3] https://tenthamendmentcenter.com/2016/06/30/what-does-the-wo...

> the colloquial use at the time of the phrase "arms" almost certainly referred to small arms

Perhaps it did, but whether it did or not, I don't think its use in the Constitution can be reasonably described as "colloquial". Since no specific restrictions were set out by the framers, I think they intended a broad right, not one restricted to small arms. Of course my opinion is an extreme outlier given current jurisprudence on this topic, to say the least. :-)

> it is obvious that--at a minimum--the interpretation of "to bear arms" would suggest that the 1986 ban is unconstitutional, as is the 1934 NFA tax act that required registration.

I agree.

> I'm not convinced "to bear arms" includes other weapons that could not be "beared," such as warships

Currently that question is moot in a practical sense, since nobody appears to be in the market for privately owned warships. And I will cheerfully admit that if we could get to a point where it was generally accepted that people had a right to bear small arms, and the arguments were over other categories, I think that in itself would be huge progress from where we are today. If that happened, I wouldn't spend a lot of time complaining that everyone recognized my right to keep and bear an AR-15 but I was being given grief about wanting to buy an aircraft carrier or a nuclear submarine. :-)

> I don't think its use in the Constitution can be reasonably described as "colloquial"

"Arms" was used in colloquial speech and legal use at the time which converged on approximately the same definition. It's late, but there are some resources on guncite that support this argument. They're all worth reading beyond this reason, mind you, and there are plenty of arguments there that disagree with mine.

Either way, it makes for good reading should you find a topic there you're not already familiar with, and even if you are, there are some interesting twists. I don't necessarily agree with all of them.

> Of course my opinion is an extreme outlier given current jurisprudence on this topic, to say the least.

Suffice it to say that I understand where you're coming from, but in my definition a broad application of "military small arms" is probably more accurate according to my understanding of what was intended. The problem is that I don't know, nor will we ever know for certain, since the militia (rather, the people) were to act as the nascent US' standing army. I rather wish this point were discussed in civics classes, but it's not. So, the younger generations are woefully unaware of history, much to no one's surprise...

Anyway, while I also wish we could undo some of the impingement on our rights by legal challenges made in the 20th century, I'm afraid that would be an uphill battle (as you also allude) given how panicked the general public is on loosening firearms restrictions thanks in large part to the unnecessarily excessive coverage of mass shootings that seem focused primarily on stoking fear.

But, if you allowed me, I'd probably rant all night about this subject which wouldn't do either of us much good by the sounds of it, other than reaching continued agreement and probably upsetting other readers.

> since nobody appears to be in the market for privately owned warships

...admittedly a shame!

> I think that in itself would be huge progress from where we are today.

Agreed. Somewhat surprised to have this conversation on HN of all places, TBH.

> ...that everyone recognized my right to keep and bear an AR-15

I live in an open carry state and sometimes exercise that right, particularly since I live in an area sometimes subject to wild animals. But while I would love to do the same with an AR, I suspect that would probably get me called in to the local sheriff's office. Not that they would care, but I would be disappointed if someone thought it apropos to waste their time.

Freedom of assembly, like the other 1st amendment freedoms, can be regulated by government so long, per prior SCOTUS decisions. They key is that restrictions must be content-neutral and can only pertain to "time, place, and manner," additionally surviving strict scrutiny (which, in plain English, means "there's no other way to do it"). This is bread-and-butter First Amendment court case stuff; any US government class that's at least halfway competent ought to have covered this.

The coronavirus bans on public gathering are fairly clearly permissible under SCOTUS precedent.

Part of the reason why lockdowns were resorted to in the first place is that Americans are, for various reasons, not likely to listen to official guidance about staying apart. See: all those panic articles about various people who knew they were at high risk and continued to go about their daily lives, or take flights or other forms of transporting masses of people, etc.

In fact, the general take is that Americans have been pretty good at following distancing rules with or without hard rules but some were slow to do so (and most government was in retrospect slow to strongly recommend doing so) but, to a first approximation, few will do so over an extended period.

So no.

This isn't even close to true. The Houston Livestock Show and Rodeo. Spring Break. Florida beaches. Evangelical churches.

Just because a majority of people do sensible things doesn't matter. The actions of the irresponsible and sociopathic minority still create the epidemic.

Let's not lose sight of the reason why lockdowns were advised in the first place.

Absent the data from widespread testing, the spread of the disease was consistent with two epidemiological models -- low-contagion/high-mortality and high-contagion/low-mortality. In the latter model, there's not much to be done to stop the disease from infecting the population at large, and not that much that needs to be done, since the vast majority of people will emerge unscathed. But in the former model, it matters tremendously -- doing nothing will slowly but surely result in vast loss of life but quarantines can stop the disease dead in its tracks.

Therefore the most cautious course of action, from a public health perspective, was to assume the former and proceed with a regime of extreme social distancing. Unfortunately doing so had dire economic consequences.

They knew within a week or two that the lockdowns had no effect. We converged right to the same level as Sweden. That they kept pushing it is insane.

What explains the difference between NY and CA if it isn't lockdowns? The difference in death per capita between the two is staggering (nearly 30x more people per capita have died in NY than CA).

Lots of things could explain it. Here are some:

- The way deaths are counted in each state

- Daily Temperature

- Comorbidity percentage difference for residents - https://covid19tracker.health.ny.gov/views/NYS-COVID19-Track...

Also population density. In NYC, a higher % of the population is touching shared surfaces such as doors and elevator buttons just to get outside.

That wouldn't affect IFR

We don't have an IFR, we have a CFR, which would be affected. Remember, all we know thus far is the numerator and a low-confidence denominator.

> - Daily Temperature

it's been unseasonably cool in socal for the duration of this so i doubt it's that.

The stay-at-home order can't itself have made a 30x difference, because New York only imposed it 3 days later than California. California started social distancing in general earlier and better, but I don't think anyone would dispute that social distancing in general matters. (In fact I think that's the obvious explanation, that due to the nature of NYC many people can't social distance to a useful degree.)

Air pollution? Well no. L.A. smog. But, there are signs higher NO and diesel particulate are a problem.

To be fair, the US engaged in lockdowns well after the situation had run its course in China so we had the data. It feels much more like a panic reaction than a question of which epidemiological model it best fit.

No way. China did an even more intense lockdown. If anything, the data from China indicated we should have done more, sooner. And that’s assuming you can even trust the data from China. Which you can’t.

Yeah, that's what I meant by ran its course. We had something like 80,000 outcomes to base a course of action on.

What if the swedes are trying to find out how to live with this virus? Like..what if there were no cure and no immunity? Unlike other governments who are betting that there will be a resolution in the near future?

Something that you might consider looking at when comparing countries or areas is the ratio of active cases to recovered. A higher ratio suggests that the epidemic is in an earlier stage.

Sweden has about 23 active cases for each one recovered. The US is about 10:1, Europe overall is about 2:1, so is Italy, and Spain is close to 1:1. Asia as a whole is about 1:1.

That likely has to do with availability of tests, who's getting tested, where and when. With Sweden not attempting to stop the spread, I doubt they've done wide-spread population testing. They're likely limiting it to hospitals, so the fact it's flat its really what matters until they kick off their wide-spread serological study in the works.

[edit] I misread, disregard, but preserving for posterity.

Active vs. recovered is a metric related to the people who are tested and confirmed. It cannot be changed by testing fewer people, only by rapidly ramping the number of tests up or down.

Ah pardon, I misread. You're totally right.

There is random sampling being done. 2,5% had an ongoing infection in Stockholm in late Mars.[1]

Now, 11/100 blood givers in Stockholm who has not been sick in the last two weeks have antibodies according to researchers.

That gives you a naive mortality rate of 0,4% per infection, assuming the spread among bloodgivers is the same as the population in general and just dividing the death count by 11% of the population. It is a high eastimate as I guess many wont get traceable antobodies and mostly asymptotic people give blood.

[1] (Swedish) https://www.folkhalsomyndigheten.se/publicerat-material/publ...

The answer to "what is the correct action to take in the face of uncertainty" is not "whatever people say we should've done after we gain full information." That's nonsense, it assumes that next time you should just know the future.

I don't understand what calculation you're doing here.

The worst flu season we saw in a decade killed (~80k / 300M) = .02% of the US population. NYC has seen ~14000 deaths, or .18% (~14k / ~8M) death rate for the entire population, and while the absolute worst may be over, people are still dying at a high rate in NYC.

The only way I can make sense of this is that you're somehow mixing figures relating to deaths in healthy populations (COVID) vs. overall death rates in flu (including older/unhealthy populations).

Nah, my calculus is basically as follows. The disease is not bad for people who aren't old, and who do not have pre-existing conditions. The data is pretty unequivocal there. The number of deaths under 20 worldwide rounds to zero. It's incredibly contagious (and contagious while no symptoms are shown), and a vaccine is 12-18 months away at a minimum. There's no way we're going to be able to contain everyone indoors, with no jobs, for 18 months while we await a vaccine that may not arrive. And that's just here in the US -- the weakest link dominates.

If we don't develop herd immunity ASAP, and instead pursue a course of lockdowns, as soon as we lift the lockdowns (either voluntarily or because people just walk out -- see the midwest), we'll immediately start playing rolling lockdown whack-a-mole as China is. The first contagious person who flies in from a foreign country (or domestically) without perfect lockdowns will re-ignite the wildfire.

We should do exactly what Sweden is doing and what the UK proposed: isolate and provide support services to those who are at risk, and let out those who aren't. I think it speaks volumes that Sweden's new infection rate stabilized at the same time as the rest of the world but without lockdowns.

It's really the only path forward. Is it perfect? Of course not. People will die. However, there's no world in which about 70% of the population won't get the disease before the vaccine arrives, so we need to control who gets it, when, and in what order, to minimize harm.

Your calculus has no numbers in it.

Different kind of calculus.

> but without lockdowns.

While I agree mostly with your assessment, it's worth mentioning that the lack of official lockdown in Sweden doesn't mean there isn't a lockdown. According to Google's mobility data for Sweden, their lockdown activity looks pretty similar to what we see in the US. People are voluntarily staying home.

> their lockdown activity

Since Sweden doesn't have a lockdown, this statement is meaningless. A major objection to lockdowns is their involuntary nature. In general, there is a big difference between choosing to stay home, and being coerced (with the threat of force) to do so. One is "choosing to do what you believe is best", the other is "prison".

And there are big differences in social cohesion between countries. Voluntary efforts appear to be sufficient in Sweden precisely because there is a high level of compliance.

In the meantime, the coastal town that I live in is getting swarmed with out of town visitors every weekend. This is in violation of the shelter-in-place orders. Other towns have stepped up enforcement to combat this type of behavior, but our police department is small and does not have the resources. We shut down the parking lots, but now they just park in the neighborhoods. And to add insult to injury, we have a large senior housing complex at the entrance point to my neighborhood.

I have little faith that the US could achieve the type of distancing and isolation necessary on a voluntary basis. There is a sharp vein of individualism that runs through our society that works strongly against us in these types of situations.

For what it's worth, Sweden's goal is to get 70% of the country infected to achieve herd immunity. They likely quite rightly believe that's the only way to stop the disease once and for all. While it's absolutely not okay to be anywhere near a seniors residence, the rest is likely tolerated because it's kind of the unstated goal to the extent healthcare facilities remain un-saturated.

It's by no means a "hope and pray it goes away" or a "lets wait it out until a vaccine" -- it's a "let's get everyone not in a risk category infected as fast as possible so long as the healthcare system retains some excess capacity."

I think almost everyone you are debating in this thread would support opening things up provided a few conditions were met. The primary condition for most of us is we need adequate testing and contract tracing capacity. And the reason that this is necessary:

> healthcare facilities remain un-saturated

You can't simply look at the current burden on the health care system to guide the process. The virus has a 2-3 week lag time between when an infection cluster breaks out, and when the health care system starts to feel the impact. Without wide scale testing, we are just going to end up back in a lock down once the infection numbers start climbing again. And I can't think of a worse scenario for our economy than having to shut things down every other month because our government is to incompetent to implement a tracing program that multiple countries already have up and running.

The serological surveys are showing 50-100x infection rates of those tested. If tested healthy persons are dying at 0.5% that gets you to 0.01%-0.005% death rate in healthy persons. That number is lower than the all cause death rates of 25-34 year olds...

https://www.cdc.gov/nchs/data/databriefs/db355-h.pdf

I see. You're correct that while the overall death rates in NYC are vastly higher than any season flu, it's mathematically possible that those are all deaths of unhealthy people.

It's a creative interpretation of the facts.

One important note: it's impossible that NYC is overestimating cases by 100x. Close to 2% of NYC has confirmed cases.

New York is at 1.3 percent positive with new cases flatlining. That fits very squarely with what you'd expect to see based on the surveys. Healthy people aren't dying by any significant number based on New York's own published data. The only other one I've seen is MA that has 97.5% deaths with comorbidity. This is a huge overreaction.

99.2% of the folks who died in Italy averaged 80.5 and had an average of 3 underlying conditions, so that data lines up.

The other thing to keep in mind with death counts, especially in NY and NJ, is that all deaths of likely-infected people are being counted as COVID deaths regardless of cause of death.

This has included a couple lf suicides of folks who had respiratory illness prior to death, and one person who got in an auto accident and died of head trauma -- but he had tested COVID-positive.

Yes, most people right now who get a respiratory illness and then die probably did have COVID -- and it is possible that some folks are dying of COVID and being uncounted to offset some of the overcounting -- but with death numbers counted so loosely, it is hard to know the real story, and impossible to do simple maths using rates from one place at one time to compute rates at other times or other places.

I tend to think that NYC must be approaching saturation, and that the true new-case numbers must be falling there, but it's impossible to answer with certainty using only the numbers we have here on the internets.

> is that all deaths of likely-infected people are being counted as COVID deaths regardless of cause of death.

No, this isn't what's happening.

Doctors who are sure to the best of their knowledge and experience that the deceased had covid-19 will put covid-19 on the death certificate, and they will also say if they think it contributed to death.

That's not the same as "anyone who dies with covid-19 is being described as killed by covid-19".

I actually followed up with you on this one recently. This is in fact what's happening, and it varies by country/region/city to what extent.

In some countries they absolutely do count anyone who dies while in the possession of COVID as a COVID death, for instance Italy. " Italy’s death rate might also be higher because of how fatalities are recorded. In Italy, all those who die in hospitals with Coronavirus are included in the death counts."

“On re-evaluation by the National Institute of Health, only 12 per cent of death certificates have shown a direct causality from coronavirus, while 88% patients who have died have at least one pre-morbidity – many had two or three.” [1]

In New York they're counting speculative COVID deaths of anyone with respiratory illness even if they've never tested positive [2].

"A subtler issue is what to do when the patient has other serious medical conditions. If the person suffered from chronic lung disease, then became infected with the virus and died of pneumonia, the immediate or primary cause would be pneumonia as a result of COVID-19. The lung disease would be listed as a contributing condition, said Sally S. Aiken, president of the National Association of Medical Examiners." [3]

The CDC has guidance on this but it's fair to say its interpretation will vary from place to place. "COVID-19 should be reported on the death certificate for all decedents where the disease caused or is assumed to have caused or contributed to death" -- that's pretty broad. [4]

Sorry, Dan, this is looking like it's not the Ebola infection you're making it out to be.

[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

[2] https://www.cnbc.com/2020/04/15/coronavirus-new-york-city-st...

[3] https://www.inquirer.com/health/coronavirus/coronavirus-covi...

[4] https://www.cdc.gov/nchs/data/nvss/coronavirus/Alert-2-New-I...

You write this long post, but then finish with the CDC guidance which agree with what I said -- that doctors have to use be able to say to the best of their knowledge and experience that the deceased had covid-19 and that it caused death, or that it contributed to death.

The problem here is that you don't know what "COVID-19 should be reported on the death certificate for all decedents where the disease caused or is assumed to have caused or contributed to death" means.

Try to find some 5 year / 10 year averages to see how excess mortality because of covid-19 compares.

Here's one for the UK: https://twitter.com/ONS/status/1252518098047041536

It's not clear that 100% of excess mortality is COVID, or how that actually relates to its IFR. More data needed.

Sure, all that excess mortality is likely to be an increase in cake-related death caused by quarantined shut-ins eating themselves to death.

I don't think it's right to compare whole US population with worst-case NY population. Of course worst-case would be much worse that average case.

Depends on what you're comparing. I agree we can't assume that the rest of the country will see similar infection rates. However, it's less clear to me that fatality rates would be incorrect.

On the other hand, there's one very specific way that considering NYC gives us data we can't (yet) get elsewhere. While it's possible that we're underestimating disease prevalence by 50x elsewhere, it's basically impossible in NYC (almost 2% of people already have tested positive, with less than 10% of the population being tested).

> However, it's less clear to me that fatality rates would be incorrect.

Depends on whether mortality rates can be variant on environmental conditions, such as viral load, population density, access to healthcare/ICUs, population socio-demographic profile, etc.

These are good points, and I think I ended up claiming a much stronger claim than I should've. There's plenty of reasons NYC could have a higher death rate than elsewhere. I don't know how precisely you could estimate the potential death rate from just NYC's data.

What I would still say is that it's reasonable to say "NYC has .18% of its population dead. It's extremely implausible that this disease is no more lethal than the flu is, when that's 10x the overall death rate from the flu. If you present me evidence suggesting that from elsewhere, it's going to have to be quite strong to overcome the evidence from NYC."

the flu would be way more deadly than it is if there wasn't a seasonal flu shot that gives herd immunity.

edit: it shouldn't be controversial to say that a thing that kills 30-80k per year with active measures of mitigation already something that would be more deadly without a vaccine. seriously reflect on that. the corona virus is something we need to take 100% seriously but also acknowledge that the flu is also very deadly... even more so without any mitigation (thankfully we have for the common flu)

The flu probably wouldn't be dramatically worse if we didn't have a vaccine. The vaccine isn't particularly effective all things considered, ranging from 10% to 60% depending on the year and how well scientists were able to guess which strain would be predominant.

The flu as-is causes 45,000,000 sicknesses every year in the US alone.

Note that the test NYC is using only tells you whether the person has Covid-19 at that point in time, so it's entirely possible to test even 100% of the population and still substantially underestimate disease prevalence if you tested most of them at the wrong time (which you probably would!)

What exactly do you mean by worst-case here? The actual worst case would be if nobody took any precautions, which almost certainly would result in more than 0.18% of the total US population dying.

New York seems to be affected by COVID pandemics way worse than any other place in the US. That's what I mean by "worst case".

Way worse, today.

Let's see. NY has 965 death per million. Next state is New Jersey (which is basically "New York by another name" in this situation) with 493. Next is Connecticut (see above) with 372. Next is Louisiana - which had a grand idea of proceeding with Mardi Gras - with 285. Do you think things will get three times as bad in Louisiana than they are now, while not getting any worse in New York? Or which state would you say would be the next worst case, not today but in the future? I'd certainly bet NY (with adjacent NJ and CT) will keep this sad championship.

> Do you think things will get three times as bad in Louisiana than they are now

If people there stopped all preventative measures, yes it would get much worse than 3x.

About 1500 people in the US died yesterday of covid; year-to-date, about 1300 people died each day of the flu.

So it seems like there should be no possible way to frame it as less severe than the flu, even only comparing to when the flu is at its worst and even assuming 100% of the population has covid or whatever other extreme things you want.

That's not entirely true. The flu has a vaccine (of varying efficacy -- 10-60% depending on the year). The flu also has a large body of immunity built up across the population from having had it. It's also less contagious.

If we assume that COVID is spreading like wildfire (it is), and that we're front-loading the cases (we probably are), and that immunity will be built up as one-off instead of like having the disease mutate substantially every year like the flu, over a few years, it could easily be better.

I'm not saying it is or isn't, just that there is a path.

Note that for the flu, most of the community does literally nothing. I mean, some people get vaccinated, but most literally ignore the flu as if it didn't exist - except if they get sick of course, and even then some people go to work if they can (they shouldn't, really, but they do).

If we compare this with complete shutdown of all public activity on the other end of the spectrum, we can see there are things we could be doing that are on neither ends. Maybe distancing but not shutdown, or avoiding some mass gathering but still being ok with individual meetings, or having places like restaurants operating at reduced capacity to ensure people aren't too close to each other, or ask people to wear masks when they enter stores, etc. And strict lockdown for places like nursing homes, but more relaxed for places like college, or having people of high-risk groups to work from home, but people with lower risk be allowed to work in the office. I'm not saying it's the right way to do right now, but it's a possibility, there are options.

So if this is e.g. 50 times worse than the flu, then trying any of the above may be too risky. But if it's kinda sorta like the flu, maybe somewhat worse but not 50x worse - then it may be prudent to consider measured response, given that for the flu we basically have no response at all and we're ok with that.

So, given that I was quoting global figures for the flu, on an apples-to-apples basis it seems like covid is about 10x the flu at the moment, in the US, ignoring the fact that it could be much worse without the lockdowns.

None of us know for sure, but isn't it plausible on the face of it that something 10x worse that could easily become 100x or 1000x worse warrants the lockdowns? I don't know if this is common sense, but in an alternate universe it could be.

> So, given that I was quoting global figures for the flu, on an apples-to-apples basis it seems like covid is about 10x the flu at the moment, in the US, ignoring the fact that it could be much worse without the lockdowns.

At about 10 weeks into the H1N1 influenza pandemic, we had a CFR of between 0.1 and 5.1% depending on the country as compared to 0.07 to 14.93% for COVID. H1N1 ended up with an actual IFR of 0.02% or one fifth of the low end of the CFR range.

Honestly, it's probably a maximum of three times as bad as H1N1, especially when you consider Italy later admitted of its 13.22% CFR, 88% of that number was "people who happened to die while COVID positive" instead of "likely died of COVID". [1] Huge deltas in CFR are likely attributable to mess-ups along the way and how COVID deaths are counted.

10X as bad is likely on the very highest end as Sweden hasn't locked down anyone, suggesting common sense instead, and their new case rate flattened out right alongside the rest of the world. 1000X is totally out of proportion with the data. [2]

I suppose we don't know for sure but we do have two and a half million data points which is enough to make a pretty good guess.

It's kinda starting us in the face that it's kinda like a bad flu except worse for old people, who we should lock down and look after as we build herd immunity.

[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

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

This can’t possibly be true about the flu. There’s been over 100 days this year and nowhere near 130000 people in the US have died of the flu in any year in recent memory

Good catch, I think I was using global figures from worldometer.

aka lying

> year-to-date, about 1300 people died each day of the flu.

That seems impossible. CDC estimates 61k flu deaths in 2017-2018 and 34k in 2019-2020: https://www.cdc.gov/flu/about/burden/index.html

That is nowhere near 1.3k per day.

You are right, my point would have been much stronger with the correct figures.

How many healthy people do you think that dies because of the flu?

Probably the same number of young and healthy people that die of COVID. For instance COVIDs fatality rate is lower than that of the flu for all demographics of people under the age of 35 per CDC data. It's about the same for 35-44 and really only diverges over 45. Worldwide the fatality rate under 20 rounds to zero. To date, pneumonia far outstrips COVID. [1]

[1] https://www.cdc.gov/nchs/nvss/vsrr/covid19/index.htm

This is completely inaccurate, and nowhere in the linked document is there supporting evidence to your claim that the flu has a higher fatality rate for under 35, nor the same for 35-44.

And this report relies on figures that diverge dramatically from the COVID death tolls. It lists the current total report as around 15k, when we're close to 3x.

> This is completely inaccurate, and nowhere in the linked document is there supporting evidence to your claim that the flu has a higher fatality rate for under 35, nor the same for 35-44.

Here's some more data. What I'm saying (that it's approximately the same as the flu for the young, with some early evidence pointing to it being better) isn't particularly controversial and well supported by evidence [1] (also older, April 14th). CDC as of late March "... no ICU admissions or deaths were reported among persons aged ≤19 years. Similar to reports from other countries, this finding suggests that the risk for serious disease and death from COVID-19 is higher in older age groups." [2] In Italy nobody under 30 died, and only 5 people between 30 and 39, as of April 19th -- yes in spite of a CFR of what, 13.22%? [3, 4]

More data as of April 20th from Oxford [4]. Scroll through all the data the world has to offer, I'm not wrong on this. I repeat, the facts you're arguing with me over, without citing data to support your case are not controversial. Surprising maybe, but not controvercial.

Especially when you consider (an albeit bad flu): the overall case fatality rate as of 16 July 2009 (10 weeks after the first international alert) with pandemic H1N1 influenza varied from 0.1% to 5.1% depending on the country. [4] The flu can absolutely be very bad is the take away there. And yeah this is worse than the flu, by around 3X on both ends of that range. Not bad, not great.

> It lists the current total report as around 15k, when we're close to 3x.

They actually address that critique, that it takes a few weeks for case data to be finalized and reported upwards, so the CDC data can be up to a few weeks behind. It's not divergent, it's delayed. This is a live list updated and maintained by the CDC. Make of that what you will.

[1] https://www1.nyc.gov/assets/doh/downloads/pdf/imm/covid-19-d...

[2] https://www.cdc.gov/mmwr/volumes/69/wr/mm6912e2.htm

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

[4] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

> What I'm saying (that it's approximately the same as the flu for the young, with some early evidence pointing to it being better) isn't particularly controversial and well supported by evidence

There are 400 covid-19 deaths reported so far in the 18-44 age group in NYC. There are 3.5 mn people in that age group in NYC. That's over 11 per 100'000.

The mortality per 100'000 for the 18-49 group in the US estimated by the CDC was in the last nine flu seasons: 1.8, 2, 1, 1.2, 0.7, 2.5, 1.5, 0.5, 3.9.

It's a stretch to say they are approximately the same and definitely 11 is not better than 0.5-3.9.

I would call 4 vs 11 "approximately the same" for all intents and purposes when the denominator is so huge especially when COVID deaths are much more generously assigned than flu deaths, per my sources, especially [4].

Not to mention, flu deaths are attenuated by flu shots, and pre-existing immunity. It's totally plausible that there are many more COVID cases than flu cases in that age group -- and of course those flu deaths will happen year after year while COVID is a very stable virus, and if you get it once, you probably won't get it again.

Certainly not enough data to conclude it's way out of line with the flu for this age group, in this season let alone if you factor in a few seasons end on end.

Lastly, with H1N1, the numbers are quite different, too.

The COVID-19 infection fatality rate in the 18-45 group is at least 0.11% in NYC. And that's assuming that everyone single person has been infected, that no-one else is going to die from now on and that the estimates are not going to be revised upwards because of under-reporting.

The infection fatality rate for seasonal flu [edit: in the slightly older 18-49 group] is around 0.2% considering symptomatic cases, and probably there are as many asymptomatic cases which give 0.1%.

So yes, it's not impossible for the infection fatality rate to be similar. With some strong assumptions including that it's five times as contagious. So yes, it's not impossible for the acumulated lethality over five years to be similar. And if you extend the period the common flu will be much more dangerous, specially as this young people became older.

> In Italy nobody under 30 died, and only 5 people between 30 and 39, as of April 19th

7 under 30 and 40 between 30 and 39. One order of magnitude more.

https://www.epicentro.iss.it/en/coronavirus/bollettino/Repor...

My data was older, and that difference is utterly irrelevant with a denominator so large, and there are many potential explanations. Especially since Italy, per my sources [4] in the parent post, assigns anyone who died while in possession of COVID as a COVID death. In fact they later announced up to 88% of their COVID deaths likely weren't actually COVID deaths but deaths of someone who happened to have COVID. So if we lop off 88%, well, it's hardly out of line.

> In fact they later announced up to 88% of their COVID deaths likely weren't actually COVID deaths

I wonder if you can back that up with a reference.

88% of CFR potentially not being COVID was from [1] referencing [2]. Specifically:

"The way in which we code deaths in our country is very generous in the sense that all the people who die in hospitals with the coronavirus are deemed to be dying of the coronavirus.

"On re-evaluation by the National Institute of Health, only 12 per cent of death certificates have shown a direct causality from coronavirus, while 88 per cent of patients who have died have at least one pre-morbidity – many had two or three," he says.

"Other experts have also expressed scepticism about the available data."

"Report from the Italian National Institute of Health: analysed 355 fatalities and found only three patients (0.8%) had no prior medical conditions. See Table 1 in the paper; (99% who died had one pre-existing health condition): 49% had three or more health conditions; 26% had two other ‘pathologies’, and 25% had one." [2]

(For what it's worth in my reply I called it out as "[4] from the parent post" which is [1] here -- sorry for the confusion -- all I meant was that the Italian data skews very high, both because it's the oldest region [Lombardy] in the oldest country in Europe [Italy] -- and because they were very generous in how they ascribed cause of death).

[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

[2] https://www.stuff.co.nz/national/health/coronavirus/12044372...

That's from one month ago and it doesn't mean that those people would be dead equally if they had not been infected.

Actually the official number is grossly under-reporting COVID-19 deaths:

"We estimate that the number of COVID-19 deaths in Italy is 52,000 ± 2000 as of April 18 2020, more than a factor of 2 higher than the official number."

https://www.medrxiv.org/content/10.1101/2020.04.15.20067074v...

Time will tell, I suppose!

In the weeks from March 1 to April 4 there were 19824 people death in a subset of municipalities in Lombardia where data is already available, while in the last five years the number of deaths in the comparable period was in the 6767 - 7248 range. This subset normally accounts for 73% of the deaths in the region so we can estimate that there were 17000 excess deaths (27200 vs 9200-9900).

Less than 9000 COVID-19 deaths were reported in Lombardia by that time. If you think they are too generous classifying deaths as being caused by the infection, what would you say that caused the death of more than twice the usual number of people during the period?

> Underlying illnesses include Diabetes, Lung Disease, Cancer, Immunodeficiency, Heart Disease, Hypertension, Asthma, Kidney Disease, and GI/Liver Disease

That's a whole lot of people left out of your number. A quick google suggests that 1/3 of Americans have hypertension and 9% have diabetes. It seems like probably the majority of the US has at least one of these underlying illnesses.

What is CFR?

"In epidemiology, a case fatality rate (CFR) — sometimes called case fatality risk — is the proportion of deaths from a certain disease compared to the total number of people diagnosed with the disease for a certain period of time."

As compared to the IFR: "The term infection fatality rate (IFR) also applies to infectious disease outbreaks, and represents the proportion of deaths among all the infected individuals. It is closely related to the CFR, but attempts to additionally account for all asymptomatic and undiagnosed infections." [1]

One's the risk that given you show up at a hospital, you die. The other is the risk given that you get infected, you die.

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

Case Fatality Rate; the rate of people who are confirmed to have COVID-19, and die. The problem is that the people most likely to be tested are those in the worst condition.

I think the obvious conclusion (just in logical progression if you take this premise as true) is that the percentage of NYC residents who contracted it was much higher than previously assumed.

Yes but the difference in fatalities is so large that it would imply majority of New Yorkers were infected and the city has achieved herd immunity. It’s not implausible just find it hard to believe.

It's hard to believe, I would think, because it would be such a remarkably positive development.

But just from a logical perspective it follows clearly from this data, and there's nothing fundamental about the situation that makes it particularly unlikely.

Of course it's quite possible that the LA data isn't really applicable for any number of reasons. But the idea that COVID will run through dense populations mercilessly, but quickly, and then sort of peter out is completely within the range of possible and non-shocking outcomes.

It's the natural outcome of an extremely communicable disease with a large prevalence of asymptomatic transmission and there's plenty of corroborating evidence for the idea that this is one of those.

It's also quite likely that the false-positive rate with this test is super high. Who knows.

If 50% of NYC has had the virus, they won't have herd immunity yet -- of these studies are correct, then the virus is horrendously infectious -- making 90%+ of population required for immunity.

It needs to be emphasized here:

This virus is much more transmissible than the viruses we usually deal with in humans.

If a virus has a low fatality rate, but is extremely contagious and prevalent, hospitals will get overwhelmed. A small percentage of a big enough number is itself a big number.

The subway and high-utilization of buses are likely factors in NYC vs. LA. LA is much more representative of the typical US city.

Frankly, I'm not sure why so many on HN are surprised by this. The scientific data on asymptomatic carriers coming out of Italy, Germany, South Korea, and Iceland has been very clear on a majority asymptomatic frame for this virus. I think that maybe authorities and responsible people are concerned that this information could lead to people flouting lockdowns more.

Even the Diamond Princess showed a 50+% asymptomatic rate and that group skewed way, way older than average -- you know, people who take cruises.

50? Just initially, but in a week or two more of these developed symptoms, so more recent number was just 18%:

https://www.cebm.net/covid-19/covid-19-what-proportion-are-a...

"Setting: Diamond Princess cruise ship, Yokohama, Japan (n=-634 tested positive).

Proportion: 18%"

Also:

"23 Residents of a Long-Term Care Nursing Facility King County, Washington

10 (43%) had symptoms, and 13 (57%) were asymptomatic."

But then:

"Seven days after testing, 10 of 13 asymptomatics developed symptoms"

That means that just 3 of 23 remained asymptomatic there too, i.e. 13%.

It is estimated that the people mostly spread this virus exactly in the phase where they still haven't developed symptoms, and the virus can be detected. But that doesn't mean they will remain without the symptoms.

I suspect most people notice a bit of a cold or "allergies"

Like you said, probably only 10% to 15% truly asymptomatic.

Thanks, I actually didn't realize the number drifted so much over time. Good to know! I'm curious how many of them were actually seriously afflicted vs. a stuffy nose. I'd argue almost 20% actually asymptomatic is still huge in context.

Especially as we know the disease affects these demographics up to 100X more than the young.

For instance consider: "Emerging evidence suggests many more people are infected than tested. In Vo Italy, at the time the first symptomatic case was diagnosed, about 3%, had already been infected – most were completely asymptomatic." [1]

[1] https://www.cebm.net/covid-19/global-covid-19-case-fatality-...

> a stuffy nose.

Not a covid-19 symptom.

EDIT: I'm wrong, sorry. Thank you to kgwgk for the correction.

https://www.who.int/news-room/q-a-detail/q-a-coronaviruses

The most common symptoms of COVID-19 are fever, tiredness, and dry cough. Some patients may have aches and pains, NASAL CONGESTION, runny nose, sore throat or diarrhea.

I mean NYC publishes the number of positive tests each day and it’s dropped from 60% to 30%. So majority of people proactively getting a test because they are symptomatic are showing up negative. Hard to believe a very high percentage of all people have had Covid, then. https://twitter.com/natesilver538/status/1252317247785271297...

Isn't that the virus test, not the anti-body test? Maybe a fairly low percentage have the virus but a high percentage have been exposed and have the anti-body.

The optimistic take on this is: that means soon enough will be exposed that their immunity will stop the spread.

The pessimistic take on this is: we also don't have strong evidence that the previously exposed are immune, ie, have a strong enough immune response to stop the virus.

You are conflating a test for active virus versus an antibody test. my suspicion is that a very large percentage of people in certain neighborhoods of New York City are going to show up having been exposed and never presented symptoms. I just don't compute how else you can explain this with a virus that is so transmissible. we should also keep in mind that the number of deaths isn't necessarily indicative by itself of number of infections since deaths can be caused by hospitals being overrun.

This only tells if you are sick not if you were sick

Those 30% numbers are still very high, the similar rate in Italy has dropped bellow 10% for 2 weeks now (it's the line-chart on this page [1] bellow "Tamponi giornalieri e contagiati"), with the latest week's figures somewhere between 5% and 7%.

[1] https://lab24.ilsole24ore.com/coronavirus/

Norway's number of positive tests was 5% at the current peak of the epidemic, for reference. Testing and isolation measures have been very effective here, so we have not had an overwhelming wave of infections (yet, knock on wood).

It's rough because it's simultaneously very mild in the vast majority of people, but a horrific, suffocating death in a tiny minority, and highly contagious.

I don't know if there is a "right" policy answer on what to do with something like that.

As someone else said, the lockdowns are probably illegal, you can't just throw away freedom of association over the flu. But if we didn't have them, we might have seen another 100k dead..

". . . the lockdowns are probably illegal, you can't just throw away freedom of association over the flu."

Where did you study constutional law? Rights granted in the constitution are subject to many restrictions. There is no explicit "freedom of association" in the Constitution, although it has been held to be implicit in freedom of speech. Free speech doctrine includes a "time, place, and manner" test that's applied to restrictions on speech when the government has a "significant interest" in the restriction[1] (e.g., national emergency for a viral pandemic). I don't know how the lockdowns (which are mostly not full lockdowns) would fare if Supreme Court were to apply the time, place, manner test, but I wouldn't presume to offer an offhand opinion that they're "probably illegal".

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

Not op. I am not a lawyer and I’ve not studied constitutional law. What I recall from high school is that anything not expressly stated as illegal is legal for citizens and any power not expressly given to a state or federal government is unavailable to them.

A law has to exist that gives states and the federal government the ability to do what they are doing and that law needs to not violate special protections built into the Bill of Rights.

With a state of emergency declared, some things can temporarily be enacted that under normal circumstances would not be legal.

I don’t know if all the lock down rules are legal, but I wouldn’t be surprised to find out that governments have overstepped their bounds.

You may be remembering something, but you seem to be remembering it almost exactly backwards. It sounds like you are remembering something about the 10th Amendment, which reserves to the states any power that is not granted to the federal government in the constitution: https://en.wikipedia.org/wiki/Tenth_Amendment_to_the_United_...

What part is backwards? I'll state what I did in reverse (ie, what I believe to be incorrect)

A) Everything is illegal for a citizen unless granted as legal by a state or federal govt.

B) The fed govt. can do anything it wants. The only things it cannot do are things that are expressly forbidden.

C) A state can do anything it wants unless a law prevents it.

D) A state of emergency does not expand the powers of the the state or fed govt.

Those are the reverse of what I said I remember and none of those sound right.

(for C/D, ignoring the supremacy clause)

It is C that is pretty close to true. Laws don't give states their powers. It is in fact the states that make their own laws, subject to some restrictions from the U.S. Constitution.

there's a really big pro-density crowd online which is working hard to undermine this argument in order to satisfy its agenda.

>Frankly, I'm not sure why so many on HN are surprised by this. The scientific data on asymptomatic carriers coming out of Italy, Germany, South Korea, and Iceland has been very clear on a majority asymptomatic frame for this virus. I think that maybe authorities and responsible people are concerned that this information could lead to people flouting lockdowns more.

Because people on here panicked without thinking and don't want to admit it.

No one wants to say "Oh yeah, I ignored second year calculus and never bothered plotting logistic growth curves with the full spread of expected fatalities. Because of this I supported legislation that's started the next Great Depression. But I meant well!"

We don't know that the fatality rate is off, we just suspect and hope it. Measurements of this kind are how we'll find out.

Are you claiming that everything is normal and would benefit from a complete reopen? Are you also claiming that the economy is in otherwise healthy shape regardless of the interest rate levels and global buyback programs that have been ongoing prior to the outbreak?

A complete reopen, no - crowded stadiums and amusement parks probably shouldn't happen. Things would never have been normal right now, and there's no silver bullet to make them normal again.

I can't speak for the original commenter, but I think the evidence increasingly suggests that we would benefit from allowing people to visit small groups of friends and run non-essential businesses.

Already 0.1% of NYC's population has died from COVID-19 - 10k deaths of 8.54 million people. That's... quite a bit higher than LA's current 576 of 10 million.

So either everyone in New York City was infected, or most probably 200-400k infected in LA is an order of magnitude of wishful thinking (e.g. ~2% false positive test rate.)

Not from, with. Also, gov’t pays a hospital out $70k for every covid death.

This "with, not from" stuff collapses when you look at 5 year averages.

EG this from UK https://twitter.com/ONS/status/1252518098047041536

Another option: A bunch of people will die in a few weeks.

Could some of the discrepancy in regional death rates be explained by a mutation in the virus?

Honestly, it seems more likely that California's much maligned car culture has served us well in this case - ensuring many of us maintain social distance which is impossible on NYC's subway and bus system. All the positive press about governor Cuomo's leadership and handling of the crisis seems a little weird given his failure to shut-down the subways (or at least curtail a but the absolutely most urgent ridership).

> Honestly, it seems more likely that California's much maligned car culture has served us well in this case - ensuring many of us maintain social distance which is impossible on NYC's subway and bus system.

California local and state government being ahead of NY in issuing shelter-in-place orders even though NY was being harder hit even before the divergent response also makes a difference. But, yeah, the hyperdensity of the NYC metro area is a big factor.

It could.

Previous studies suggested that the strain circulating in NYC is of European origin, whereas the California strain, arriving earlier, was of Asian origin.

It's always possible that the NYC one is more fatal. We don't have data, but it is possible.

Thing is, though, at the peak, NYC was seeing a 60% positive rate on tests and is still seeing something near 30%. That pretty much guaranteed that the vast, vast majority of NY cases were not detected, and still probably are not.

(Italy is back around 5-6% again, for example)

It must be the case that a dramatic percentage of NYC is infected. Certainly not more than 100% as some oversimplified maths suggest, and likely not 80%+ that would be very comforting -- but quite likely between 30% and 45%, given the Stanford and LA studies, and the NYC testing data.

So like the fact that most of NYC, like most of Italy, and Spain is infected because they are a more dense city than LA is lying about the data because this is backed up by data in other countries like Korea, Germany and everywhere else with similar data.

There's the theory of Vitamin-D deficiency. It would make some sense given NYC's climate and time of year. In LA, people likely are exposed to the sun year round. It would be interesting to see if other respiratory illnesses exhibit a similar pattern between these two cities, as I understand Vitamin-D levels have an affect those as well.

But also, NYC is considerably more dense, so diseases spread far more rapidly. Especially so considering heavy public transit use versus, what I would say is, extreme overuse of personal vehicles.

Seems so obvious that this was the flu but far more contagious from the beginning because of the lack of herd immunity and vaccines. We are just getting verification.

"so obvious".

I love it when people have strong opinions about a complex, constantly changing piece of difficult science.

It's not opinion is the logical conclusion from literally every random sampling study. Pretty much consistently 50x as many people have been infected that we thought. And if we made the current strategy with limited data why aren't we updating what we are doing with more data.

Might NYC have more vulnerable people, or more confined spaces for spreading (elevators in high rises, e.g.) ?

https://www.worldometers.info/coronavirus/country/us/ right now shows 12850 cases per 1M in NY, which is 1.2%. Which is pretty close to what the study shows for LA. So it's not actually suggesting NY has much more than it's already known, if we assumed overall percentage would be similar for both. It does suggest California - which currently lists 853/mln - undercounts it by more than 10x. Of course, provided that the study is true.

Also, having antibodies and being hospitalized is a very different thing. If a lot of the cases are asymptomatic or very light you won't see hospitalizations and you'd never know the person had an encounter with the particular virus until you do mass testing. And NY is quite different from LA, as far as I know, in terms of how dense population can be and how much people use mass-transportation and other high-risk scenarios, and the probability of being hospitalized may also depend on that (in other words, it may matter not only whether you had any contact with the virus, but in which circumstances and how much of the contact you had).

That doesn't prove the study is true of course, but I don't see anything from comparison with NY that makes it implausible on its face.

LA is far less dense, fewer travel by public transport, and the weather is much warmer.

The people are probably healthier too.

Just speculating here but I hung out with some NYC folks last month in SoCal and they practically all smoked. It was very strange, I rarely encounter smokers living in California.

Cigarette smoking rate is widely studied; no reason to use anecdotal data here.

The tobacco smoking rate in New York is 14.1%; it's 11.3% in California. Some difference, yes, but far less difference than you see between either and some rural southern states where the numbers can be roughly double that.

https://www.cdc.gov/statesystem/cigaretteuseadult.html

Case fatality rate is # of deaths / # of positive tests. It’s enormously biased by the availability of testing, and the allocation of the tests we have to the sickest people. It’s not especially likely to generalize to the population at large. We need large scale random sampling of the population for that.

Everyone is eager for these results, because there’s a good chance that the true case count is higher by anywhere from 2x to 100x. If it is, then proportionally fewer people are expected to die as the virus eventually does churn through the whole population.

It’s probably still too early to tell, but this is promising.

No the very obvious conclusion is that covid-19 isn't very deadly. Thats what every test has concluded. Its that the death rate is closer to .3% than the 3% that the WHO quoted and much closer to the flu's death rate of .1%

Premier is an importer for "Hangzhou Biotest Biotech Co". Most of the validation data in the Stanford preprint also appear in this sell sheet:

http://img.bj.wezhan.cn/content/sitefiles/2005911/images/132...

Stanford reports doing a 30-person validation test with pre-COVID-19 hip surgery patients (100% negative). But this is not enough to validate these population-scale tests, especially when 98.5% specificity (rather than 99.5%) would fully invalidate the headline result.

> Stanford reports doing a 30-person validation test with pre-COVID-19 hip surgery patients (100% negative)

Hmm - I guess I should now go to the source - but from the other link: " test validation included a total of 30+371 pre-covid blood tests, and only 399 of them came back negative." https://statmodeling.stat.columbia.edu/2020/04/19/fatal-flaw... - I think I trust it more

Yes, the 371 are listed on the sell sheet (see tweet above). Unclear if Stanford was involved, but I would suspect not.

This one's from a different company (BioMedomics) and it was a random test of people pulled straight off the streets. Happened at an entirely different geographical area (Chelsea, Massachusetts) as well. I don't want to be too optimistic but there are some signs that we are heavily under-counting the actual number of cases (at least in the US).

https://www.bostonglobe.com/2020/04/17/business/nearly-third...

According to the Q&A it is the same test as the Santa Clara study, so concerns about false positives apply equally to both.

It seems like they should be doing multiple tests, as independent as possible, per blood sample, and seeing if they agree?

> Another interesting aspect of the result was 6% of men showed antibodies but only 2% of women tested.

That is interesting. So far, men are more likely to die of covid-19 (almost a 2:1 ratio) than women; if this is a repeatable result it would seem to suggest that the reason is that fewer women are becoming infected, not that women are somehow less likely to exhibit extreme symptoms when infected.

Or that women are less likely to need antibodies to clear this infection. Not everyone develops antibodies for every disease, the immune system is super complicated.

The Imperial College model was also published by press release. That will have ended up doing a whole lot more damage than any of these serosurveys, when all is said and done.

Aren't these results consistent with Netherland's findings (3% of blood donors had COVID-19 antibodies) ?

> Shady stuff

If I'm being honest, whether it's logical or not I wouldn't be surprised if this is politics and misinformation starting to creep it's way into science. In current events there's a struggle to open up the economy with Trump saying we're ready and the that governors are at fault for keeping everything closed while the are governors saying it's way too early and they need to keep everything closed. This may be an attempt to make the public think the situation is safer than it is (if more people have been sick than thought then we could potentially have herd immunity or be closer to it) which the public can then put pressure on the governors.

Who knows, but I will say it's getting harder and harder to say some thing are too tin foil hat lately.

No tin foil hat required. This is exactly what they are trying to do[1].

[1] https://www.commondreams.org/views/2020/04/20/trumps-reelect...

Our initial policy was based on early data from China which seemed to imply 3% mortality, an R0 of 2.4 and a 5% hospitalization rate. In the past month, data from all sources and methods is showing this to be far less deadly but also far more contagious. Perhaps it’s time to adjust our view of the lockdown.

Well, R0 is probably still about 2.4 in most cities in the United States (NYC, like Wuhan, considerably higher).

Source suggesting this: https://www.medrxiv.org/content/10.1101/2020.04.12.20062943v...

I think it has been pretty obvious that IFR is not 3% for a long time. Imperial College had dropped it to 0.67% in China 3 weeks ago (https://www.thelancet.com/journals/laninf/article/PIIS1473-3...)

I think the view that the infection fatality rate is significantly lower than the case fatality rate has already been factored into public policy.

Because even if the IFR is around 0.5% then that still works out to > 1,000,000 people dead in the US in an uncontrolled spread scenario.

It would have to be another order of magnitude lower than that to be comparable to the seasonal flu.

Uncontrolled? No, but we do know who is most vulnerable and we should be able to make more targeted and surgical measures to limit the negative impacts of social distancing.

Reply on news.ycombinator.com