The death toll for Covid-19 in Nembro, Italy | Hacker News Reader

The death toll for Covid-19 in Nembro, Italy

corriere.it

110 pointsmyth_drannon134 comments

There a few possible (and reasonable) explanations in the HN comments, one more is that often covid-19 leads to pneumonia and that may be what is attributed as the cause of death, (nefarious or not).

I did hear rumors of Japan doing this intentionally to make it look more under control than it was.

Another possibility is that medical resources are being monopolized by covid patients, or people don't want to go to hospital for fear of catching it themselves. This situation is still caused by covid but wouldn't be shown in the reported deaths from covid.

The article brings up a valid concern, but doesn't explore any other possibile explainations which is irresponsible when people are already panicking.

The numbers really don't seem consistent across all the countries. Just one example is the death rate in Germany is 0.9% while the death rate in Spain is 8.5%, nearly a magnitude of difference with both countries have several 10s of thousands of cases, and that number varies between the two extremes all over the continent (and world).

There are multiple variables at play. For example in Italy there is a high percentage of people living with their parents in the age 20-40s bracket, which should lead to more elderly contamination. Also the health services between Germany and Italy should not be even comparable in terms of equipment and capacity.

Do you think that's enough to account for a 146 to 6 difference in deaths per million between Spain and Germany?

In Italy the hospitals were overwhelmed so you should expect the mortality to increase very fast when you can't even give basic treatment to patients in need. The same thing happened in Wuhan and that led to massive death toll. But as I said, many variables at play and we will surely have a better understand of what mattered the most to prevent mortality.

After you run out of ventilators the sky is the limit when it comes to deaths.

When you still have ventilators, only a few very elderly/unlucky die...

So a country with 10,000 cases and 1,000 ventilators can have 20 deaths, and another with 10,000 cases and 500 ventilators might have 400 -- if of the 10,000 cases in both 520 need ventilator... (and some die, whether with ventilator or not, but many wouldn't if they had ICU/ventilator treatment).

Medical system capacity and utilization explain the vast majority of the variation in mortality among different countries, after accounting for stage of the epidemic.

When things are not above capacity, mortality is 0.5% or less. After reaching capacity, mortality jumps 10x or more.

Age distribution can only account for variations within, at most, a factor of two.

So Spain's (and France's to a lesser degree) medical system got overwhelmed that much faster than Germany's or the UK's?

They might have been very late with testing. I know this was barely on the radar in France even 2-3 weeks ago. MRSA has also been suggested as a component.

You'll need to back up the claim that "only a few very elderly/unlucky die"

What % of people actually survive after being on a ventilator? Because I've seen various studies and anecdotes of this for C19 specifically, with results ranging from "basically no-one" to "80%, at least temporarily, but hardly anyone is actually out of hospital yet so maybe they're going to die later".

Beyond C19, this study suggests roughly something like 50% long-term survival IIUC: https://www.ncbi.nlm.nih.gov/pubmed/26003390

Ventilators are invasive things that themselves can do lung damage. Fairly common for people to choose not to go on them in end-of-life directives.

Mortality from ARDS is 50% even with ventilators. Source: wife is a pulmonologist and I am a radiologist.

Are there other reasons why C19 patients would need ventilators? Or is ARDS the main reason?

Is that mortality rate consistent across different underlying causes?

ARDS is the main reason. 40% mortality from just ARDS. Mortality goes up with covid19, age, sepsis, and renal failure.

> health services between Germany and Italy should not be even comparable in terms of equipment and capacity.

Italy's health care is among the best in the world, and Lombardy's is several points above the Italian average.

Germany has apparently 2.6 times the amount of IC beds available in proportion to the population- in the Covid epidemic this should offset the reach of maximum capacity by a week.

You can't compare those numbers without an in-depth understanding and comparison of who is being tested in each country.

Which means the numbers are not consistent.

I'd just say the numbers are different.

I like to thing of it as a unit problem. The numbers are presented as x "items", but the definition of "items" for each of the numbers is not consistent (or, even, actually known).

Another case in which the numbers are oddly (one might say "suspiciously") different are the UK's number of critical patients (163) compared to their known number (19522), which is very low at 0.8% compared to France's 11.5%. That's a massive difference in the number of seriously critical patients, given that the UK has a sufficiently large sample size now.

> critical patients

What does that even mean? The UK isn't testing unless you need hospital care (or you're special, e.g. Prince Charles, Boris Johnson, Idris Elba), aren't all of those patients somewhat critical? Or is "critical" just "hooked up to a ventilator" in which case it could be limited by the number of ventilators instead?

I read that Germany is not testing the dead, so if someone wasn’t tested while alive then their death is attributed elsewhere. Italy is testing the dead to get a more accurate picture. I find it baffling that there isn’t an international standard for testing in a situation like this so that we can get a handle on what is actually going on. If we don’t have data, how can we possibly draw inferences and learn anything?

There are international standards and the standard is that flu is not normally considered a cause of death. It's not realistic to expect autopsies to be performed and coroners reports to be written on every person who dies of pneumonia during an epidemic.

And yet I understand this is what is happening in Italy where the suspected cause of death is Covid19. If we don’t make it realistic to expect this (during a pandemic), then how can we trust the numbers?

Right, for example what if there is a second strain going around that is obscured by the large inconsistencies in testing and reporting?

Wildly different case fatality rates might be caused by the same infection fatality rate if milder infections are systematically less likely to turn into medical cases in some countries.

Compare the age distribution of infections in Iceland versus the Netherlands for instance.

https://twitter.com/alexandreafonso/status/12435570137597009...

We have every reason to think that just as large a fraction of the 10-19 year olds in the Netherlands have gotten this in the Netherlands as in Iceland but when it stays as a cough and light fever many medical regimes are never going to test those cases.

And countries have different demographic profiles, though not enough to explain a 10x difference. And medical systems get overwhelmed in some places which could possibly be up to a 10x difference in a really cases.

A huge factor here is time. 3 weeks ago I saw a bunch of people saying "Look, South Korea tested a lot of people, finding the majority of the mild cases, and thus only has a death rate of 0.5%". Since then, however, deaths/cases in SK has risen to 1.59%, and increasing daily. They don't have many new cases, what's happening is the cases from 3-4 weeks ago are dying.

Germany has tested a lot and found a lot of cases earlier in the disease progression. I'm sure their death rate will rise like S.K.'s did. Not all the way to the 8% - 10% we see in Spain and Italy, but I'd guess up to 2%.

> I did hear rumors of Japan doing this intentionally to make it look more under control than it was.

That is probably happening in Japan because of lack of testing (it's almost impossible to get tested) and ultimately "up to the physicians" - and as you mention it's also possible there are some strict government guidelines to make it look less serious than it actually is, but who knows...

> people don't want to go to hospital for fear of catching it themselves

To be exact, this group of deaths would be caused by the fear of COVID, as you say.

How about we at least track those deaths caused by unavailable treatment or postponed, but critical surgeries. If my community is affected I don’t care if Covid is killing us directly or indirectly. They all should be counted at least under a different category. Ignoring them would like claiming puerto rico only had 100 deaths from hurricane

The "direct/indirect" distinction is very important because once we have more perspective on this disaster, we should try to find out whether the alarm and the various extreme measures helped or not, and to what extent. I'm not saying that we should ignore the downstream effects of the pandemic, I'm saying the opposite - we should look at them with extreme scrutiny.

Exactly. That would be very helpful for the next pandemic.

>I did hear rumors of Japan doing this intentionally to make it look more under control than it was.

Tangental note, but this caught my eye. It's exactly the kind of comment I would expect to hear in a movie about a global disaster. "I hear Japan is hiding their numbers", "I hear China has a secret cure", "I hear Missouri is untouched by all of this", "I hear..."

I would appreciate it if we avoided this kind of hearsay on HN during these difficult times.

Then it would be good to explain clearly why Japan is such a large outlier while Korea was not. And please no "they wear masks" and "they wash hands" bullshit, because there are no more masks in Japan for months now and hygiene is approximative at best - and apart form closed schools it's pretty much business as usual in Japan so there is virtually no reason why the virus would not run its course here as well.

>Then it would be good to explain clearly why Japan is such a large outlier

I decline. It might be what coopsmgoops suggests, or it may be something else. I don't posess the facts and therefore I won't speculate on matters of public health.

Japanese ppl don’t like touching each other they don’t really shake hands socially but they bow when saying hello and good bye. That is not the main reason but things like this in a culture adds up. Japanese is also exceptionally clean

I would appreciate if you can breakdown how cleanliness in Japan (say vs. the US) helps stop (propagate for the US) the spread of the Corona Virus. I have seen the Japan is exceptionally clean argument which is independent of the Japanese don't like to touch each other and are less touchy-feely as a culture (hugs, handshakes, etc. ) than Western countries.

> Japanese ppl don’t like touching each other they don’t really shake hands socially but they bow when saying hello and good bye

Yet they stick to each other like sardines in a can when they take public transport every single day in Tokyo and Osaka - and we are talking about dozens of millions of people who transit every single day. That should be the primary mode of exposure especially since masks have completely run out in Japan for months.

So, unlikely to be a good explanation, sorry.

This might just be a transient anomaly, perhaps exacerbated by insufficient testing.

But if it's a real difference, the possibility of genetic resistance ought to be explored. It's not that it would likely make any practical difference. But it would at least be interesting, and one never knows.

Korea is also a large outlier. It would be even more of an outlier if not for megachurches which don't exist in Japan.

China has a history of embellishing and fabricating all sorts of published economic numbers. The number of funeral urns being collected is raising questions about the true toll (https://time.com/5811222/wuhan-coronavirus-death-toll/). Democratic countries can skew their ICU or death statistics, but only so far.

Iirc, trump has censorship powers under the emergency act. If he really wants business to reopen after Easter, I hope journalists are going to be ready for some fact checking

By "censorship", do you mean no numbers or fake numbers?

There is one possible explanation: China annual death rate is around 7.3 per 1000. In a city with 11 millions people (wuhan), there would be 6600 death per month without COVID 19.

> I would appreciate it if we avoided this kind of hearsay on HN during these difficult times.

This graph makes it quite clear that Japan is hiding their numbers (also Iran and South Korea):

https://aatishb.com/covidtrends/?country=Australia&country=C...

The Graph also shows that China either has a cure, or is fuzzing their numbers as well.

At minimum, China does not include asymptomatic cases in official counts.

https://www.bloomberg.com/news/articles/2020-03-24/china-s-h...

Well, China, in particular, is a known quantity here. They reported a total of 144 influenza deaths in 2018 of 700k infected. Personally, I consider that a lie. Others would say that they simply use different statistical methods. The truth is, they use this sort of number fudging internally as propaganda. See, the west suffers hundreds of thousands of influenza deaths per year, right? And China only had 144. Clearly TCM is superior to Western medicine, etc...

> I did hear rumors of Japan doing this intentionally to make it look more under control than it was.

Japan, Iran, South Korea, and maybe China are all fuzzing their numbers:

https://aatishb.com/covidtrends/?country=Australia&country=C...

Notice how they are the outliers compared to a ton of countries.

South Korea has tested more citizens than anywhere else on Earth. That's for a population of 50 million.

Are you sure they are "fuzzing" their numbers or a matter of confirmation bias here?

I think when all the dust will settle, we will know the real death count by doing a delta of total deaths of 2020 vs. 2019 and 2018. There will be two main reasons of the extra deaths:

1. Deaths related directly by Covid-19. (either the disease itself, or the underlying conditions got worse by the disease)

2. People that didn't have the coronavirus, but died because they couldn't get the usual care due to hospitals being totally overwhelmed (things like heart attacks, strokes, car accidents, insuline/asma shock, etc... turn into death sentences)

Same as in war, while many deaths are caused directly by the fighting, a lot of the other extra deaths are indirect (famine, disease... etc)/

Also the article assumes 100% infected. If it is only 50%, then death rate is 0.5%. Although the graph does show there are more deaths expected.

We can actually expect the mortality rate is a lot lower than 1% if the hospitals are not overloaded.

Counterpoint as at March 23rd: “Among 3,711 Diamond Princess passengers and crew, 712 (19.2%) had positive test results for SARS-CoV-2 (Figure 1). Of these, 331 (46.5%) were asymptomatic at the time of testing. Among 381 symptomatic patients, 37 (9.7%) required intensive care, and nine (1.3%) died (8). As of March 13, among 428 U.S. passengers and crew, 107 (25.0%) had positive test results for COVID-19; 11 U.S. passengers remain hospitalized in Japan (median age = 75 years), including seven in serious condition (median age = 76 years).” - https://www.cdc.gov/mmwr/volumes/69/wr/mm6912e3.htm

So mortality rate is 1.3% and expected to rise, but that is probably skewed by age cohorts, so rate in countries with more young people than a cruise ship might be expected to be lower.

https://www.euromomo.eu/outputs/zscore_country_total.html

If I am reading that correctly, some countries like Spain see almost annual death spikes to around 8 standard deviations. That doesn’t make sense, what did I miss?

Potentially a 2020 vs previous years approach would even slightly underestimate the death toll.

I have to assume being in lockdown suppresses other 'normal' causes of death (fewer car accidents, or conventional disease spread). Depending on how long these lockdowns really last, it could actually dent the real impact of those causes of death, making the covid peak look slightly smaller vs the average.

We can subtract traffic accidents, homicides, suicides, falls etc and compare only remaining numbers.

That the official numbers underreport Covid-19 deaths by a factor of 4 is one hypothesis -- but it's also possible that this is how many people would be saved by quick interventions at the hospital, but who otherwise died.

For example, maybe the ICU beds were full for a month, and in that time they would've saved 96 people, but couldn't.

Are you saying a spike in non-covid deaths — e.g., heart-attack, etc? That’s certainly a possibility and one of the main dangers at the moment with an overwhelmed healthcare system. It also raises an interesting question of how you account for the toll of the disease: on one hand you have deaths due to the disease itself, and on the other deaths that could have been preventable had the disease not overwhelmed hospitals.

Depending on how you count, you could hypothetically end up with mortality rates >100%. I.e. a disease that kills everyone it infects, and also closes down a hospital causing deaths among people it doesn't infect.

Good point. Some part of the increase is almost certainly this ... just hard/impossible to know how much.

Edit: actually, i guess if you also looked at the top 10 typical causes of death, seems like you could account for it (ie, heart attack, MVA, or whatever would show abnormally low numbers).

All the statistics seem like a mess for this - we know every single region is doing different amounts of testing with many only testing highly symptomatic people, yet all these are being combined together. Then different protocols for what counts as different levels of severity etc. It will be interesting when people have had more time to sort through the data and clean it up properly what the real estimates of mortality and transmission actually are. Right now everything seems plausible right through from 50% of the population already had it and is immune through to the death rate being drastically under-estimated and we are about to experience an apocolyptic event for humanity ....

Excellent work, and I hope more people in other cities start running these calculations.

> In the hypothesis - not at all remote - that all citizens of Nembro have caught the virus (with many asymptomatic, therefore), 158 deaths would equate to a lethality rate of 1%. That is precisely the expected and measured lethality rate on the Diamond Princess cruise ship and - made proportionally by demographic structure - in South Korea. We have made exactly the same calculation for the municipalities of Cernusco sul Naviglio (Mi) and Pesaro using exactly the same methodology. In Cernusco the number of anomalous deaths is equal to 6.1 times those officially attributed to Covid-19, also in Pesaro 6.1 times. But even more staggering are the Bergamo figures, where the ratio reaches 10.4.

This seems an unreasonable assumption, no? I don’t think any models predict a 100% infection rate: herd immunity would kick in beforehand, at the least.

If true, then the death counts and death rates would be higher. E.g. with 50% infection, deaths are 4x, 12x and 20x, and the death rate is 2%.

Of course, you also have to account for local demographics.

Edit: I checked the numbers for Bergamo, and they’re troubling. It’s a city of 122,000, and it has a 0.5% death rate for the whole city.

Further, people are still sick and dying, so these death rates will only go up.

Shouldn't the Covid-19 be immune to herd-immunity though because of the very long incubation time, during which you can still transmit the virus?

Herd immunity kicks in when a large fraction of the people you encounter were already infected, giving you fewer chances to infect new people. So incubation time doesn’t matter there (though certainly it doesn’t help, for other reasons.)

> Herd immunity kicks in when a large fraction of the people you encounter were already infected

Alternatively, immunized people.

I think there is a momentum effect if the percentage of infectious people gets high enough. Where you overshoot the static herd-immunity threshold. If you can keep the number of infected below 10% then you don't see it. At least that's what I saw when I tried to model it. (Take that as you will).

Naturally-occurring herd immunity is a temporary effect. Only artifically-imposed herd immunity (i.e. vaccination) is lasting.

Yes, but unless herd immunity lasts only a couple weeks that shouldn’t be a factor in these Italian analyses. The infection spread over a short period.

Some diseases become endemic and pervasive in childhood and end up not causing issues in the elderly who already built up their immune system in childhood. Other times an exposure in childhood ends up cross reacting or something and makes things more deadly later on (I think that's one theory for the 1918 flu; elderly and middle age were protected from an earlier flu exposure, 20-30 year olds had exposure in childhood to something different with a shared antigen that ended up making the immune system overreactive and made things more deadly).

I don't know why this comment is being voted down when it is, in fact, correct. Though "temporary" is on a scale proportional to a lifespan.

If an epidemic sweeps through and infects most of the "herd", that herd has herd immunity.

In time as individuals born and others die, the fraction of the herd that is immune goes down and the herd is vulnerable to the same epidemic.

But with vaccination, herd immunity can be sustained indefinitely. With no further epidemics. But, as measles has shown, only as long as a high enough fraction of the herd vaccinates.

Of course it is also possible that if enough of the herd is immune for long enough that the epidemic goes extinct. This takes a lot of work but is how we got rid of Smallpox

Effective contact tracing and social distancing can achieve the same end, which is how we got rid of SARS.

IIRC the level at which herd immunity effects start kicking in is dependent on how contagious a given virus is. For an entire village to be infected in less than a month requires a very, very high R0 (one that's higher than has been reported so far), which would mean that herd immunity would likely kick in at >0.9.

I don't think an R0 that high is plausible, which suggests that only a fraction of the population has been infected in Nembro. That, unfortunately, suggests an even higher mortality rate, at least when the medical system is overloaded. The decline would then be due to the effectiveness of quarantine measures.

Fully agree: 100% infection for everyone is very unlikely, everything considered.

how is this excellent work? it's unscientific conjecture based on eyeballing some extrapolated data

It’s excellent in the sense that no one has done this sort of analysis yet. We’re flying in the dark due to constraints of testing.

The hope would be that by publishing this work, more people in other cities are prompted to analyze excess deaths and note whether there are any abnormalities during the time the pandemic hit locally.

Wasn't there a thing were deaths were recorded for the place where the person died, not where they normally live? So maybe those people just died in a hospital in another city? (I don't remember if this was for Italy, but I remember reading that about some place in connection with Corona).

In any case, why not make an effort to reach out to the cities in question and ask for an explanation?

Has anyone seen any data on how this has made the average daily deaths in the US rise? I know that number probably isn’t updated real time like the Covid-19 numbers. I think that would really show how much this is affecting us if the numbers go up drastically over the expected amount of deaths per year.

I have found the averages from last year, but haven’t seen any more up to date data on that.

https://www.cdc.gov/nchs/fastats/deaths.htm

I know of deaths at the local hospital ER with overlapping symptoms but not being officially counted as the deceased were never tested. On a county level there’s maybe one or two recorded. This testing debacle is hurting everybody in more ways than one.

Epidemic Modeling 101: Or why your CoVID19 exponential fits are wrong

“Over the past few weeks, a terrible affliction has been spreading across the world. Otherwise healthy and productive members of society have been infected with this devastating illness that causes them to fire up Excel, Python or R and start extrapolating the latest numbers of confirmed CoVID19 cases in their town, state, country or even the world!”

https://medium.com/data-for-science/epidemic-modeling-101-or...

I remember the death toll for H1N1 to have been reported in the thousands when it happened but recently read that after a few years they said it was more like 150,000. It's possible they were just reporting US deaths where the 150k number is worldwide. How come the response for H1N1 is nowhere near as intense as the response for COVID?

It's possible that 150K was a number reported by somebody else, with a motivation for reporting a very high number. I'm not saying I'm convinced the number is lower, but just as much as there are entities that under-report deaths for selfish reasons, there are those that over-report them.

Based on current trends, Covid-19 is trending to be around 10x or more worse than the 150k figure. Current projections have the US alone at 100k-200k deaths.

That would be why this response is more intense.

Except the high estimate for H1N1 in 2009 from the CDC is 575,000 deaths worldwide.

https://en.m.wikipedia.org/wiki/2009_swine_flu_pandemic

> The Infectious Diseases Society of America estimated that the total number of deaths in the U.S. was 12,469

The seasonal flu is a pandemic every year and kills a similar amount of people. It’s a shame people don’t take it seriously enough to get vaccinated.

H1N1 is a newer variant of seasonal flu. But it’s not nearly as lethal and it doesn’t usually leave survivors with the kind of lung damage COVID-19 does.

It’s about carrying capacity of the healthcare system. Flu already strains hospitals, but COVID-19 is multiple times worse so people go without treatment because here aren’t enough beds or doctors and nurses. That’s the difference.

Every time I look into the flu I'm reminded just how big of a deal it is. The 2017-2018 season was particularly bad in the US, but I don't really recall a ton of press about it. Some stats from the CDC for the US alone:

“CDC estimates that the burden of illness during the 2017–2018 season was high with an estimated 45 million people getting sick with influenza, 21 million people going to a health care provider, 810,000 hospitalizations, and 61,000 deaths from influenza.” (emphasis mine)

https://www.cdc.gov/flu/about/burden-averted/2017-2018.htm

Will be interesting to see what all of the social distancing/quarantine does for the flu. The map at healthweather.us has been pretty interesting to watch. Personally I think they just screwed up the 'atypical' visualization by stretching it back to March 1st rather than showing present, but if you click on any county and scroll down you will see that they are almost all below standard observed illness for this time of year.

https://healthweather.us/

Cars kill almost as many people as influenza, and costs far more in marginal QALY (younger, healthier people).

Flu isn't so incredible. What's insidious is that it's just not bad enough to make it worth the cost of prevention (and it's hard to prevent -- the vaccine is a luck shot every year, as flu has strains and mutates), but it's the rumble before the storm, as one of a family of viruses of varying parameters that occasionally strike far worse (every few decades).

A general solution to viruses, something like gene therapy, would be a very big deal, perhaps more than cancer treatment, in eliminating perhaps the largest non-self-imposed threat to humanity.

> We’ve updated our atypical Illness map to reflect the cumulative amount of atypical illnesses we’ve observed since March 1 — previously this map reflected only new atypical illness, updated daily.

Whyyyy??? that's incredibly stupid way of obscuring information by multicounting, that at best adds no value because the integral of an exponential is still exponential, and at worst makes a hash of it. The only rationale I can see is to make the numbers bigger and more clickbaity.

Moderators: is it possible to have "(in Italy)" added to the title?

It's not clear this can be extrapolated to other countries. (In fact, they reference South Korea as a benchmark for the true value.)

   Nembro should have had - under normal conditions - about 35 deaths. 158 people were registered dead this year by the municipal offices. That is 123 more than the average.

  In the hypothesis - not at all remote - that all citizens of Nembro have caught the virus (with many asymptomatic, therefore), 158 deaths would equate to a lethality rate of 1%. That is precisely the expected and measured lethality rate on the Diamond Princess cruise ship

The author appears to be using the total dead (158) rather than the excess dead (123) for this calculation. That suggests that rather than the 1% "exact match", the lethality rate should be 0.78%.

Good catch. Though I looked up the population of Nembro. Wikipedia has it at 11,518.

So would the rate actually be a bit over 1%? Bergamo’s was about 0.5% by the same metric incidentally.

Good ideas here of why it could be happening.

Let me add another: fear and hopeless. These kind of emotions may make people more prone to get sick with other illnesses and depress the immune system so it does not fight back as it really could.

Some references:

- https://www.medicalnewstoday.com/amp/articles/324090

- https://psychcentral.com/news/2014/05/22/study-probes-how-em...

- https://www.simmsmanncenter.ucla.edu/center_events/healing-a...

So if the mortality rate of the decease is what it is in Germany or Korea - about 0.5% but the number of dead is 4x the official 10.000. Then about 8 mio - ca 15% - of Italy’s population has had it. Is that really correct?

Korea's CFR is 1.6%. Germany is 0.9%. Both are climbing over time as the time lag from detection to death is on the order of weeks.

Korea's used to be, what, .6% when they stabilized the number of infections? At the time I was hoping the true IFR was that low but as you say this thing can take a long time to kill people and given what a good job Korea's done with staying on top of their testing I'm not holding out hope that the IFR is below 1%.

It could be, but then the R0 should be greater than 3. Which is kind of scary.

Here in the UK, I personally know a lot of people who suspect that they have it, but aren't in official numbers because their symptoms are mild and access to tests is poor. 15% doesn't seem beyond the realm of possibility.

Beyond a certain point I think it's actually less scary it's more people have it, no? Because it means that the serious cases are a smaller proportion.

But that also means it will spread faster and peak higher, which creates a larger burden on hospitals.

Spread faster, yes. But if enough people have it who are not being included in the stats, then I think that means it will peak lower, as it would mean that the percentage of people who need hospital care would be lower than is being assumed.

I know several people in Norway who suspected they had it, got tested and returned negative. Based on this anecdotal information, I wouldn't put too much stock in those kinds of guesses without testing.

Yeah, that's a good point. Hopefully they'll get testing rolled out here soon!

it must be more complicated because it's very unevenly distributed - Milan was spared, and it's only 3km from bergamo

Oops, 45km

The stats for this are all over the place. Speaking for the UK, no distinction at all is made between died of cononavirus and died with coronavirus. We've had a University of Oxford study claiming that 50% of us have already been infected (a figure so outlandish you might wonder if it was rather a University in Oxford) in which case we should expect the died with figures to be a great deal more already, and what's more that herd immunity should be taking the pressure of the system any week now. But I can't help but feel this belief in hidden large denominator is wishful thinking.

50% does seem outlandish, but from what I can see, the numbers infected are a lot lot higher than officially reported. I would go as far as to say that the majority of cases aren't being reported, because almost nobody without severe symptoms has access to testing.

I'd be curious how much variation there is in previous years; are there any previous significant month spikes that have been washed out by the average?

It doesn't seem plausible that random chance could account for 2020's, though.

I don't know what the distribution of deaths looks like normally, but a 4x difference is going to be several sigma unlikely in any reasonable one.

Also, consider that it happened exactly at the same time coronavirus deaths spiked, which seems vanishingly unlikely to be a coincidence.

Problems attributing deaths to COVID-19 will probably be common worldwide. Radio Free America had this similar report about excess deaths in Wuhan being far out of line from the confirmed COVID-19 deaths:

https://www.rfa.org/english/news/china/wuhan-deaths-03272020...

(Note: This article is much more speculative and weakly-sourced than the Italian one. Also, be aware that RFA is funded and overseen by the US government.)

It is possible that more people than usual are dying from other causes (heart disease, cancer, injuries) because the overwhelmed medical system is unable to treat them.

If so, it's still indirectly caused by the pandemic.

but not an actionable statistic for other countries

Is anybody familiar with the authors or the website? These findings are shocking if true, but how do we verify their numbers or decide if the source is reliable?

Good one. Always compare the current death rates with the death rates from the prior year or two. That will explain anomalies in reports by various countries.

I believe in asian countries where sickness(which can spread) is stigmatized(i definitely believe it is in India..), people will be hesitant to get tested/treatment. There has been several news reports of medic professionals being asked to vacate rental homes, miscreants calling out names in front of patient's home. Not sure why they do them. This will just cause people with symptoms to not get tested.

Can anyone actually tell me what was the cause of death if an 85 yo already had 3 diseases prior to this virus?

If a person has a terminal disease and has N months to live, but dies sooner from a car accident, the net deaths did not increase. This person was on the death queue.

edit: bring the silent downvotes. Also, a number of my family have tested positive for this virus.

Pretty important to capture the terminal catalyst to help understand the lethality of a pathogen.

"They would have died anyway" is true of anyone and tends to mask critically important information.

It’s true of everyone given 150 years. Sure, but an 85 yo with 3 diseases was on the front of the queue.

We need to test for those who had it and didn’t die. All evidence is pointing at orders of magnitude that had it but didn’t show

Let me get this straight (all of these are straight from the article):

1) Nembro, in the province of Bergamo, is the municipality most affected by Covid-19 in relation to the population.

2) hypothesis - not at all remote - [is] that all citizens of Nembro have caught the virus ... Given the decline seen in the last few days after the peak, flock immunity has likely been attained in Nembro.

3) 158 deaths would equate to a lethality rate of 1%. That is precisely the expected and measured lethality rate ...

So, the worst hit city in the worst hit region, with collapsed medical system and elderly population; the absolute upper bound on possible deaths; and we get 1% until herd immunity? This, again, is the upper bound of ultimate worst case, based on the article.

Isn't this a huge revision DOWN from the previous 2.5-3-6-8-15% numbers people are throwing around?

EDIT: what's up with the silent downvotes? I'm quoting the central number and assumptions directly from the article and asking a question. Should I take it as "you are right but I dislike it"?

it's bad science to pick out one small sample of the most affected town (especially when arguably they could have considered the entirety of deaths) and extrapolating that "death rate is 4 times" . There is no statistical testing here, no confidence intervals, nothing. The 4x number is unsubstantiated. The hypothesis that the numbers are misreported seems to be true, but statistical tests are missing.

I think there are already a few good probable explanations here. However without more clear data it is hard to find the real drivers.

Has anyone looked at the equivalent numbers in the US? Or South Korea, or other places with outbreaks?

Was in the news that in China the funeral homes urn deliveries are suspiciously larger than expected. In Russia there is a spike of pneumonia looking like dicease and deaths...

The unexpected number of urns answered by the quote from the article that initiated that “news”: There were 56,007 cremations in Wuhan in the fourth quarter of last year.

After a month or two of lockdown in Wuhan, the number of backlogged urns is tens of thousands, that is to be expected.

Now incorrect: Bloomberg lowered number from 56,007 to 13,856 in their article — see note at bottom of https://www.bloomberg.com/news/articles/2020-03-27/stacks-of... which says “(Corrects characterization of 2018 and 2019 cremations data in story published March 27.)”

Original pre-correction here: https://archive.md/g1BDx#selection-2503.0-2507.171

I wonder why no one is using Wuhan, China numbers. Not reliable?

The same phenomenon has been anecdotally reported there as well, with a lot of untested "pneumonia" deaths: https://www.scmp.com/news/china/society/article/3050311/its-...

Evey time I see an article critical of Chines government in SCMP I feel a shock until I remember it's a Hong Kong paper. But then I feel a little confused over Hong Kong's semi autonomous status.

This is beyond a spurious claim.

[Retracted]

No. In four areas in Italy, including a mid sized city.

> We have made exactly the same calculation for the municipalities of Cernusco sul Naviglio (Mi) and Pesaro using exactly the same methodology. In Cernusco the number of anomalous deaths is equal to 6.1 times those officially attributed to Covid-19, also in Pesaro 6.1 times. But even more staggering are the Bergamo figures, where the ratio reaches 10.4.

Just wait until the deaths of despair hit from the crashed economy and isolation. Alcoholism, opiates.

Counterpoint: While suicide increased, life expectancy overall increased during the Great Depression:

https://www.pnas.org/content/106/41/17290

Reply on news.ycombinator.com