Hacker News new | past | comments | ask | show | jobs | submit login
SARS-CoV-2 seroprevalence in healthy blood donors during Covid-19 Milan outbreak (medrxiv.org)
58 points by panta 11 days ago | hide | past | web | favorite | 64 comments

This study [1] published by Policlinico di Milano found that 4.6% of blood donors had anti-SARS-CoV-2 IgM/IgG antibodies at the start of the pandemic (February 24th).

See also these articles (in Italian):

- https://www.policlinico.mi.it/news/2020-05-20/1660/covid-19-... - https://www.lastampa.it/cronaca/2020/05/20/news/il-coronavir...

[1]: https://www.medrxiv.org/content/10.1101/2020.05.11.20098442v...

> 4.6% of blood donors had anti-SARS-CoV-2 IgM/IgG antibodies at the start of the pandemic

Which would make sense. Inter Milano's player Romelu Lukaku was saying about a month ago that the vast majority of his teammates (23 out of 25) had already gotten the virus in January, and that for the Inter - Cagliari match (which was played on January 26th) one of his teammates had to be substituted off after only 25 minutes because he couldn't run anymore.

For whatever reasons Inter's officials didn't like his comments and reprimanded him, but it seems like Lukaku was closer to the truth than many would have liked to admit a month ago.

> For whatever reasons

Perhaps because he was making baseless claims about his teammates and tainting the reputation of the club. He should at least know that players being taken down by cold or flu is not unusual in football.

2014 Seven players are struck by 'flu symptoms': https://www.telegraph.co.uk/sport/football/teams/germany/109...

2017 Wenger fighting to contain outbreak: https://www.dailymail.co.uk/sport/football/article-4296828/N...

Feb 2019 'Man flu'[0] sweeps through League Two club: https://www.bbc.com/sport/football/47239534

March 2019 'epidemic' with eight players: https://www.thesun.co.uk/sport/football/8709002/poland-flu-e...

[0] https://en.wikipedia.org/wiki/Man_flu

> He should at least know that players being taken down by cold or flu is not unusual in football.

One could think so, of course, but then how does one explain the fact that when testing really started in Italy (end of February - actually early March for footballers, after the Juve - Inter match played on March 9th) not one football player from Inter tested positive, even though they were in what was even by then the center of the pandemic in Italy, Milano?

By contrast two Juve players (Dybala and Matuidi) tested positive immediately after March 9th, plus some other players from Firenze and Genoa. One of the only reasonable explanations would be that most (if not all) Internazionale players had already gotten the virus by then and had managed to get rid of it.

And I've heard of that "tainted" argument and for the life of me I couldn't understand it, being sick (or having been sick) doesn't taint anyone. It somehow reminds me of the infamous #milanononsiferma hashtag used by Milano's mayor (among others) at the end of February, people who were thinking that ignoring actual real stuff for fear of not "tainting" the city (in the mayor's case) would somehow make things better.

Did the players test positive for Cov2 antibodies?

Start is bad. February is two months after back computed patient zero is deemed to have existed in China. This is "infection demonstrated to exist in Italy before prior dates we believe"

And to an extent it's not even that. The samples were taken from the 24th Feb. Italy had its first localised lockdowns on 21st, at which point the disease had been around for long enough for a couple of identified cases to have died. The 'at the start' of the outbreak figure is for the weeks between this and the whole of the north of the country being shut down on 8th March. The figure is higher than expected [with a wide confidence interval] and is in Milan rather than the villages initially isolated, but we know initial testing capacity wasn't there, transmission can be asymptomatic and severe symptoms lag.

Prior to about the 21st of Feb Italy only reported three cases of coronavirus in the entire country with the most recent case being a Wuhan evacuee at the start of the month: https://uk.reuters.com/article/uk-china-health-italy/first-i... As you say they started to lock down very soon after that. February 24th is very much at the start of the detected outbreak, and it seems safe to conclude that the virus was indeed already circulating widely at that point. They went from basically no reported cases to the most cases in the world outside China and their healthcare system being overwhelmed in days.

While other experts had pointed out that Italy likely had more cases than already reported, the numbers they threw around were more in the hundreds range rather than substantial percentages of the entire population: https://www.nytimes.com/2020/03/21/world/europe/italy-corona... This does explain a lot though, both about what happened in Italy and why containment efforts in other Western countries failed whilst ones closer to China geographically seem to have done much better.

> This is "infection demonstrated to exist in Italy before prior dates we believe"

That is an unlikely hypothesis, given that we didn't see an exponential explosion in deaths, and people going to the hospital until February.


1. The virus was for some reason less lethal before February.

2. Only people who don't show symptoms were catching it.

3. The serological studies are mostly finding false positives.

My money's on #3.

It's also possible that people happened to develop antibodies for a different coronavirus that are also effective against SARS-CoV-2-- meaning they're true positives (effective antibodies) that don't indicate an infection with SARS-CoV-2.

This happened with someone that survived SARS-CoV having effective antibodies against CoV-2: https://www.nature.com/articles/s41586-020-2349-y

Why not the fact that perhaps the virus found its way in more vulnerable populations, such as hospitals and nursing care homes (44% of the current deaths)?

If that happened, the population would be very largely skewed towards the most vulnerable (you usually don't go to ER or a hospital if you're healthy) and that would explain the uptick in deaths.

There's no other idea I can offer to explain why a disease with an estimated IFR between 0.5 and 0.9% would cause an order of magnitude more deaths.

For that 44% number, what data set are you using?

I have been idly tracking the data here in WA. 90% of the deaths have been in 60+ years of age. Just ~30% percent of the cases. In that age range, the CFR is staggeringly high. I don't know the data for the numbers from nursing homes.

Sorry, forgot to mention it. These are the numbers from Italy's Istituto Superiore di Sanità.

No worries. Just curious if there was a global source for that.

From looking at the numbers, I can't help but think we really failed protecting the elder population.

>That is an unlikely hypothesis, given that we didn't see an exponential explosion in deaths

The unmitigated doubling time of COVID-19 seems to be around 2.5 days. So if about 1% gets hospitalized and people noticed after 10 or so patients with the same symptoms (that's before any of them died, mind you) you'd need about 1000 people infected. To go from 1 infected person to 1000 takes log2(1000)=9.96 doublings. So if the doubling time is 2.5 that would be 25 days or over 3 weeks before anyone would notice.

Some of these numbers are a bit of a guess but I'm trying to show that "exponential explosion" can be very small and slow at the very beginning of an outbreak.

The variance in actual increase rate is probably way nosier with fewer subjects. I.e. in the beginning you can probably hover at a somewhat constant number of infected for a while before it takes off.

On average an infected person seems to spread it to 2-3 persons, but we don't know the variance. It could be something like 8/10 spread it to zero persons, 1 to 4 persons and 1 to 16 persons. With low number of infected that would make the increase very noisy. In fact, it has to be so, otherwise Covid19 would explode everywhere and not be more or less containable. It has to catch momentum or whatever.

There is also the potential issue that some of the hot spots are like that because they got seeded by multiple super spreading events.

As in unlucky happenstance means three infected people in a tour group show up and infect two to three dozen people each over a weekend. You go from 3 to 100 cases in two days. R instead of being 2.5 jumped to 30!

After a week just with normal spreading you have close to a 1000 people ill.

I can't find it anymore but there was an article here published a few days ago that essentially estimated whether infections with SARS-CoV-2 come from homogeneous transmission, i.e. everyone who has it infects a few people, or from highly concentrated transmission, i.e. there are a few people who infect a large number of others. The latter seems to be the case for SARS-CoV-2. This to me makes me believe 2. is correct. We know that this disease is non-lethal for the large majority of people below 40-50. It is thus not unthinkable that it was able to spread widely in those groups before finally hitting more vulnerable parts of the population all at once in March, leading to the devastating effects we observed. Additionally, in a model where everyone is equally likely to be infected, the number of deaths would reflect the percentage of people having the disease. That is why people think there is an upper limit of a few percent of infected. This however breaks down entirely if there are clusters of cases within different subpopulations.

#3 sounds both plausible and absolutely dreadful to me (I’m in Sweden where the de facto “strategy” is still herd immunity)...

#1 and #2 sound not entirely implausible to me though... Severity of disease seems to vary a lot with infection dose. Is it really that far fetched that when infection doses are very low the disease is mild, but as the virus level in the population rises and infection doses go up the disease becomes a very serious public health issue?

Does anybody know any data that supports/refutes my thinking above?

Is there evidence that infection dose matters? Last I heard, that was specifically not the case.

With what we know of the age dependence of the CFR, it is not unreasonable to think #1. In fact, until it broke into long term care facilities, it would fly under the radar fairly well.

Remember, for #2, it is not that most people stay asymptomatic. They are usually pre-symptomatic, and do have the symptoms later. Just for the majority of folks <60, they are a bad cold. If you are under 20, not even a bad one.

And from some comments down thread, #3 is as likely to be wrong with these having more false negatives. So, tough to say.

It is not #3, if we believe what they say in the study:

"The test had a 98.3% specificity and 100% sensitivity"

I.e. only 1.7% likelihood of false positives

EDIT: And no case was missed

I'm betting we didn't see an exponential explosion in death - ever. Look up the gompertz curve. It's very convenient for politicians to have everyone believe counterfactuals of unbounded exponentials.

I like #3 too. And really it is worth waiting for expert opinions rather than getting too excited - maybe people develop antibodies in response to some other cold virus that has a similar external surface to COVID-19. Who knows? The world is large and there is a lot going on. These 98 and 100% style test sensitivities don't magically make the world a controlled lab-like environment. I'm willing to bet crazy things sometimes happen that weren't tested in labs.

4. The current beliefs about lethality and R0 are way off

What are those beliefs and how do they have to be off to explain this and other data?

What if this "exponential explosion in deaths" is because people didn't receive medical care for various other conditions and thus deaths spiked?

This would require the other conditions to frequently enable death with COVID symptoms which wouldn't have happened with the condition being treated. For example patient has liver cancer under treatment, treatment ceases and they contract Corona-20, the untreated liver cancer then enables progression of COVID and death, whereas when liver cancer is treated then progression wouldn't have occurred.

Seems rather more complex than the common account.

We know #2 is true for younger people, as the mortality rate is almost zero unless you're in a nursing home.

Don't know why you're being downvoted.

According to Spanish government's estimates[1] based on their own seroprevalence study:

For the age group under 50, the estimated IFR is about 0.003%

For the age group 50 - 69 yo, it's 0.04%

And for those over 70, it's 4.1%

The study is considered to be well designed by Carl Bergstrom[2], for example.

[1]: https://www.mscbs.gob.es/profesionales/saludPublica/ccayes/a... Look for the table on page 15.

[2]: https://twitter.com/CT_Bergstrom/status/1261041466882678784

Please cite your data on the mortality rate.

According to Spanish government's estimates[1] based on their own seroprevalence study:

For the age group < 10 yo, the estimated IFR is 0.002%

For the age group 10 - 19 yo, the estimated IFR is 0.003%

For the age group 20 - 49 yo, the estimated IFR is 0.002%

For the age group 50 - 69 yo, the estimated IFR is 0.04%

For the age group > 70 yo, the estimated IFR is 4.1%

The study is considered to be well designed by Carl Bergstrom[2], for example.

[1]: https://www.mscbs.gob.es/profesionales/saludPublica/ccayes/a... Look for the table on page 15.

[2]: https://twitter.com/CT_Bergstrom/status/1261041466882678784

It also matches the CFR you can calculate from the Seattle Times. (Don't know if we have a good handle on IFR, but I can go off published data for CFR, right?)

op00to: I'm curious. What I said is well-known.

Were you trolling, ignorant of current news, or what exactly?

Having the antibodies in such a large percentage of the population suggests it was around for quite a long time beforehand.

It takes months to reach those levels.

In addition there's the unusually large number of pneumonia deaths in oct/nov throughout the country


We should expect to get serological data from excess Chinese mortality before December then. Assuming they agree to good back sampling. They have biopsies like any public health system.

The Chinese government has been pushing the narrative that their reported case counts exactly describe reality and there were no widespread undetected infections. I wouldn't rely on any evidence that contradicts that ever officially seeing the light of day, given the heavy central control over coronavirus-related research and the sensitivity of the subject. Supposedly there were antibody testing results from Wuhan leaked that showed a 5% infection rate there but I don't think anyone has been able to confirm this.

FTR, as the study was conducted on blood donors, it may not be representative of the whole population of the city. This suggests more than ever the need of extensive serological testing.

The whole antibody testing looks totally unreliable to me. My wife was tested COVID-19 positive about 4 weeks ago. So was her boss at a different test location, so I assume the test was correct. Also she was ill with the typical symptoms. Meanwhile she has recovered and three days ago she had and antibody test and the result was negative, not even close to the threshold.

Do you happen to know which antibody test Was used? There are a number of different ones with very different specifications.

I was a test against IgG antibodies. The report does not mention the manufacturer of the test.

You concluded that on a sample size of one?

It's anecdotal, but if you follow https://www.reddit.com/r/COVID19positive/ there are tons of similar anecdotes. In a large enough number anecdotes make a statistic.

Also, I claim that there was not even time to test the reliability of the serological tests an actual patients. So, anecdotes may be all data we have.

So it started before they thought it did.

The evidence is starting to stack up. Maybe the virus didn’t come from China.

China was just the first to discover it.

If this is true, then all the Western countries took a collective dump on China, and accused them of spreading the virus.

Some theories and timelines so far:

1) July 2019, a lot of strange pneumonia cases were reported in Virginia. The symptoms sounds suspiciously like Covid19. Two people die. [1]

2) Three months later, on October 18, 2019, French athletes that attended the Military World Games in Wuhan reported testing positive for the Covid19 antibodies. They are muzzled, and told to not talk with reporters. [2]

3) December 16, 2019, The first documented hospital admission in China.

4) China investigates, and determines this is a new virus. They report it to the W.H.O. on December 31, 2019.

5) China checks past records, and determines an earlier infection was on November 17, 2019. [3] About a month before their previous first known case.

6) A New Jersey mayor tests positive for the antibodies, and claims he got sick in November 2019, a month before China detected their first known case. [4]

7) December 27, 2019, A woman in Seattle [5], with no travel history to China, who tested positive for the antibodies, claims she got sick two days after Christmas. This is 4 days before China reports the virus to the W.H.O. And three weeks before Seattle’s first official case.

8) Jan 26, 2020, The Lancet theorizes that the virus did not originate from the seafood market. [6] Daniel Lucey asserts, “The virus came into that marketplace before it came out of that marketplace.” Thus, this was not from Chinese people eating bats.


We don’t know anything for sure, but the W.H.O. needs to investigate those Virginia nursing home residents that got sick in July 2019, and test them for the coronavirus antibodies.

If, and this is a very big if, the virus did not originate from China, then, the western nations are going to have a very big egg on their faces. Especially after all their racist attacks against the Chinese people, for eating bats, and what not. And especially for demanding reparations of trillions of dollars in damages from China. Are they going to pay up instead? The mystery continues.


[1] https://abcnews.go.com/US/respiratory-outbreak-investigated-...

[2] https://nypost.com/2020/05/07/athletes-at-world-military-gam...

[3] https://www.livescience.com/first-case-coronavirus-found.htm...

[4] https://nypost.com/2020/05/07/new-jersey-mayor-believes-he-h...

[5] https://www.seattletimes.com/seattle-news/antibody-test-resu...

[6] https://www.sciencemag.org/news/2020/01/wuhan-seafood-market...

If you’re going to post weird conspiracy theories, at least own them. Presenting a full conspiracy timeline and then saying “well we don’t know anything for sure” just makes it seem like you’re trying to avoid criticism.

The virus originated from chinese horseshoe bats. It can't just randomly teleport to some other country on the other side of the world to start an outbreak, before existing in human population of China.

Well no, but also in the US there are research facilities working with Coronaviruses [1]. I am not advocating for this theory because there is zero credible evidence for it but there are scenarios that would make this theoretically possible.

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4797993/

This is far from certain. The virus has not been identified in the wild. There are several genetically similar viruses in pangolins and bats sampled from China, but Covid combines sequences from at least 2 of them.

Did you realize that there are also bats in Austin, Texas? They live under the bridge there.

Unless these bats are perfectly clean and disease free, after millions of years of evolution. And somehow, only the bats in China, can carry a novel coronavirus?

I’m just pointing out that bats exist worldwide. And they can infect other farm animals. This is not a unique situation to only China.

So the question is what has caused increased deaths if there actually was an unusual increase compared to previous years at all? Was it something else, like a multiresistent hospital infection in nursing homes?

From the abstract:

> Conclusions: SARS-CoV-2 infection was already circulating in Milan at the outbreak start.

It just means that the virus entered the region a few weeks or months earlier than thought so far (the study started on Feb-24), not that there have been COVID-19 waves in previous years.

This seems similar to the recent news that the first case in France was backtracked to December 2019, e.g. the virus was already spreading before the first cases were discovered:


If this is true, wouldn't that mean that the R is in fact not as high as we believe it is, because it was spreading for longer on a slower rate as opposed to spreading fast in a shorter time period? If we think that it is still highly infecious then the only other conclusion can be that we have a much larger population already immunonised?

Not necessarily. r0 can be much more reliably measured than severity. hospitalization and death rates give pretty clear picture of doubling time, but could represent wildly different percents of the overall infections

Well, probably there were no previous years for SARS-CoV-2. If you are referring to the increased deaths compared to other regions in Italy, there are many factors, the most prominent probably being (a) Lombardy's terrible healthcare handling and corruption; (b) being affected much earlier than other regions they were later on the lockdown; (c) deliberate/criminal mishandling of retirement homes; (d) fewer tests performed and containment (for example Veneto has been much more proactive on this front).

I don't think it has anything to do with corruption and terrible healthcare (mistakes were made, but as a resident there, I think the system works pretty well nevertheless, or at least used to be pre-crisis).

I think it has more to do with the route of entry of the virus in the territory, which hit hospitals first (not closing down Alzano Lombardo was a grave mistake), then retirement homes (an error shared with Lazio and elsewhere).

This meant that the newly-infected population was largely skewed towards the most vulnerable, and coupled with imperfect knowledge about treatment, the net result was a lot of deaths.

As another resident (Milano), I think corruption is endemic in the regional health system. This is, of course, not to deny Lombardy healthcare compares well relative to that of most other Italian and European regions. Still, it was clearly overwhelmed and mismanaged. In my opinion, this and the timing of the first response account for the many deaths. And those in charge of both need to be held accountable.

That said, I do think the numbers of the Policlinico study might well be representative, at least regarding Milano. It is just anecdotal, but I know a lot of people there that got sick, but where unable to get tested, or waited weeks before receiving any treatment.

Unfortunately this year the flu, for those who got it, was quite strong as far as I can tell, so it's a real confounding factor.

Symptoms are quite different, though. And they are evaluated by the ‘medico di base’ (general practitioner) before requesting a viral test. So I am not clear on what could be confounding.

Exponential growth is very slow when it’s just getting started.

If that were the case, then since hardly anyone was sick, hardly anyone should have had antibodies.

Unless, of course, the antibody test is too general, and is picking up false positives, or antibodies that have nothing to do with COVID.

The media is full of news about the unreliability of antibody tests. Search "antibody tests unreliable" and you will see.

Just one example here: https://www.theguardian.com/world/2020/apr/09/uk-government-...

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact