Excess deaths as of today are maybe 50% higher than usual even at the peak, so there is no way we would able to tell that we had an outbreak from excess deaths that early on.
The only way to know would have been to perform tests, which we did not do.
All the antibody studies so far suggested infection numbers that were massively higher than confirmed cases.
Furthermore the PCR tests used for initial diagnosis have high false negative rates and most people with only mild or no symptoms never got tested in the first place.
> Excess deaths as of today are maybe 50% higher than usual even at the peak
They're still higher than usual, and so any 'even at the peak' appeal to curves acknowledges that epidemiologists were probably correct and numbers of cases rose very rapidly from near-negligible levels, before slowing due to social distancing measures. If the disease was widespread much earlier, excess hospitalisations and deaths during that period ought to be much higher than they are now with aggressive social distancing having been place for nearly two months. It's not like France has limited access to healthcare or was more likely to chalk a surge of hospital inpatients with severe respiratory symptoms off as something else in late Feb or early March than in late April, when despite lockdown they were much higher...
> It would have been invisible for a long time, spreading uncontrolled.
Except that if it really was 'spreading uncontrolled' and giving everyone 'the worst flu', as opposed to a handful of isolated cases which became more than a handful shortly after people started to panic about it, there wouldn't still be more vastly people being hospitalised for respiratory symptoms weeks after measures were taken to halt the spread than in late February. There might not have been much community testing going on then, but hospitals were certainly keeping records of who was coming in with respiratory problems; the well-established fact that comparatively few people were is strong evidence against the hypothesis it had already become widespread by then.
COVID-19 isn't giving everyone "the worst flu", certainly not to the extent that it requires a hospital visit. Most confirmed cases report mild symptoms. Without the expectation of having contracted COVID-19, getting a cold - even a relatively bad one - would not cause panic.
You're also more likely to develop pneumonia from Influenza than from COVID-19, especially if you're young. If you presented to a doctor oblivious of COVID-19 (or SARS), they would assume it is Influenza.
I'm not native english speaker and I'm wondering if this is the correct use of the word "hardly". I expect that to mean "not" or "almost not". "hardly hit" would mean "hit very little". Or am I confused?
It is that high now, it would not have been that high early on, especially considering that mortality is most likely closer to 0.5%.
It simply would not have registered among the usual deaths from viral pneumonia, which has a year over year variance of a similar magnitude.
The actual death numbers are also highly dependent on the age groups affected, in Germany for instance there is no discernible excess mortality because most of the infected are below the age of 65.
I want to believe that a bunch of us already had and got over Covid, I really do. I am in northeastern USA and a lot of people around me had this "unusually bad" flu this year. Coworkers, kids, me and family - took us a month to get over it. But, that would mean the disease was super widespread and we couldn't call that period "early on". Based on anecdata and Bayes it means a lot of people had it already in Feb, as many as during peak flu season.
This is why I believe it's just a bad flu after all - the spread pattern matches the usual seasonal flu/cold; and we would have noticed an elevated mortality.
We'll never know because we won't know how the disease would've spread without all these measures.
If we adjust for years of life lost, as opposed to just lives lost, COVID-19 may be less severe than seasonal H1N1 Influenza, which kills a lot of young people. On the other hand, COVID-19 also seems to be far more infectious.
There are several European countries that right now have no discernible excess mortality despite having hundreds of thousands of confirmed cases and thousands of deaths.
Outside of NY/NJ, the excess mortality in the US is within 20% of the normal rate, no different than the variance expected from a bad/mild flu season.
If you were to transpose those cases back to February, what would you have seen in the death statistics? Nothing suspicious at all. There could have been hundreds of thousands of cases that have gone unnoticed, because there wasn't any testing - and that's assuming a mortality rate of over 2%. If the mortality rate is lower than 0.4%, as some studies suggest, it could've been millions of cases.
You do realize we’ve been waiting it out at home since early march here in CA? The only real reason why no of deaths is not through the roof yet. If you divide number of confirmed (!) covid deaths in nyc by total pop it’s already over 0.2%
Edit: actual calc i did is 13k confirmed deaths in nyc / 70% herd immunity magic number of 8.4 mil
Let's look at Germany from that EUROMOMO map. Despite all the measures, they have had over 150k cases and over 6k deaths (a death rate of over 2%), but it doesn't show up on the graph.
All the deaths that have happened in Germany in the past weeks could've happened months earlier and it wouldn't have shown up either.
This disproves the idea that there couldn't have been any mass infections earlier, because we would've seen that from excess deaths. That's the point I am making. The only way to know would've been through testing, but there wasn't any testing then.
Hypothesis: "If there had been mass spread outside of Asia as soon as January or February, we would have been able to tell because of unusually high excess deaths"
Contradicted by: "There are known cases of mass spread that didn't result in unusually high excess deaths"
Therefore, excess death is an insufficient metric to reveal a mass spread of COVID-19 - it could have been spreading undetected.
Whether there have been any measures to limit the spread is irrelevant to that conclusion.
> "There are known cases of mass spread that didn't result in unusually high excess deaths"
You used Germany as an example.
Here's the situation in Germany:
Since 13 March, the pandemic has been managed in the protection stage as per the RKI plan, with German states mandating school and kindergarten closures, postponing academic semesters and prohibiting visits to nursing homes to protect the elderly. Two days later, borders to five neighbouring countries were closed. By 22 March, all regional governments had announced curfews or restrictions in public spaces. Throughout Germany, domestic travelling is only authorised in groups not exceeding two people unless they are from the same household. Some German states imposed further restrictions authorising people to leave their homes only for certain activities including commuting to workplaces, exercising or purchasing groceries.[10]
Tell me, please, how does this support the idea that mass spread without extreme protective measures (as was the case in January or February) can result in no excess death?
> Tell me, please, how does this support the idea that mass spread without extreme protective measures (as was the case in January or February) can result in no excess death?
I'm not saying it can result in "no excess death", I'm saying if there had been mass spread back then, even on the order of hundreds of thousands of cases, it could've gone unnoticed, because the excess death would've been within the seasonal variance.
If there are really 10x as many actual cases as reported - which is what antibody studies suggest - then the virus has either been spreading much faster than we assumed, or has been spreading for longer than we assumed.
The fact that somebody who died in France in December appears to have been infected with COVID-19 strongly suggests that there has been community spread far earlier than we assumed.
> There are several European countries that right now have no discernible excess mortality
This is because there are lags in the data. You need to wait a few more weeks (and for some countries it'll be months) for the data to come in and be reported.
I don't know the extent to which this is true, but even then, we do know the average deaths in previous years and we do know the actual deaths reported and can get an idea of high how the discrepancy could've been.
In the case of Germany, there would be 12,000 weekly deaths on average, versus about 2000 weekly deaths due to COVID-19 at the peak - that is within the variance caused by the seasonal flu. It would not have been a suspicious rise.
> In the case of Germany, there would be 12,000 weekly deaths on average, versus about 2000 weekly deaths due to COVID-19 at the peak - that is within the variance caused by the seasonal flu. It would not have been a suspicious rise.
If you're comparing covid-19 to flu you must count them using the same method, and you're not doing that here. Here, for covid-19 you're using "deaths after confirmed positive" but for flu you're using "all cause mortality". When you use the same method to count covid-19 and flu deaths you see much higher rates of death for covid-19.
> It's very easy to search for this information. Here's one link...
Your claim is that the lack in excess mortality is solely due to lag. Your link doesn't say anything about the extent of the delay regarding countries like Germany.
> If you're comparing covid-19 to flu you must count them using the same method, and you're not doing that here. Here, for covid-19 you're using "deaths after confirmed positive" but for flu you're using "all cause mortality".
The hypothesis is "If there had been mass spread back then, we would've seen it from excess deaths", which implies all-cause mortality. Of course I'm mentioning Influenza because it causes some of the seasonal variance and some of the same symptoms.
> When you use the same method to count covid-19 and flu deaths you see much higher rates of death for covid-19.
Yes, but that's in hindsight. That's not the way you would have looked at the cases at the time.
There are about 20,000 pneumonia deaths per year in Germany where the cause is never determined[1]. That's over 300 per week average, more in the winter years. Now suppose an old person comes in and dies of pneumonia. There's nothing suspicious about this. Suppose a few more come in this year than the last year. Again, nothing suspicious, some flu seasons are worse than others.
I'm not saying there have been 2000 undetected cases of COVID-19 deaths per week in Germany back in January/February, but there could have easily been 100-200.
This data is very interesting. I wonder whether it will bounce below the historical average after the peak (i.e. people dying 3-6 months earlier due to the pandemic).
Some have low specificity, but others aren't that bad. They haven't been tested to the degree that you would like to, but they have been tested. There is a margin of error of course.
Disease spread tends to always show a pretty standard shape. It was that observation (Farr's law) that led to epidemiology being born as a field. It grows, it peaks, it enters immediate decline. Importantly the peak doesn't last long - epidemics don't spend 6 months with a stable number of people getting sick at the peak. Look at graphs of COVID deaths and cases and you'll see the standard pattern.
We know now that this virus has left hospitals virtually everywhere without overload, even in places like Sweden. A few other places were hit much harder. But the response in New York (with mobile morgues) wasn't reported anywhere else. It's apparently some kind of invention of New York policymakers during a brief peak rather than a worldwide phenomenon.
A virus can be spreading and growing for a long time before it attracts attention. The evidence keeps mounting that the first reports in Wuhan were not in fact the first cases after COVID had mutated or crossed from animals but merely the first where doctors decided to search for a novel virus after coming under pressure and noticing some novel symptoms. What you saw in New York was a mix of:
1. Conditions at the absolute peak of infection, not at the start.
2. Media hype and fake news.
For (2) I present https://nypost.com/2020/04/01/cbs-admits-to-using-footage-fr... as evidence. CNBC spliced video from an ICU in Italy into reports about New York without telling anyone. Outright deliberate deception is also the tip of an iceberg: there's far more selective reporting, exaggerated anecdotes and so on. On April 6th Vox reported the entire USA was running out of sedatives needed for ventilation:
On the same day Gov Cuomo was saying they had enough ventilators with some in reserve. A few days later he was sending them to other parts of the USA.
Whatever you think you know about the situation in New York you really only have a tiny fragment of the whole picture (and the same for me and everyone else posting here). Our understand of reality lies shattered in pieces on the floor, smashed by speed, poor quality data, poor use of data, and extremely poor journalism. All we can do is work to piece together a narrative of what really happened by examining all the evidence we can get. Repeated anecdotes from many different people about having had a COVID-like illness before it was being discussed much are interesting for that reason.
Actually a personal friend who is a doctor on the frontline in Paris saw mobile morgues back in late March.
So I am capable of personally falsifying your statement:
>But the response in New York (with mobile morgues) wasn't reported anywhere else.
I lack a citation but I'm not really trying to prove anything either. I don't think New York is isolated. Lombardy seems to be stricken pretty hard, and it's hard for us to know what it was truly like in Wuhan.
Speaking for Italy: at least part of the high number of deaths might be traced back to hospital infections (such as a massive outbreak in the hospital of Alzano Lombardo, which initially was kept open after a very brief closure) which of course targeted more vulnerable people.
Also nursing care homes make up a significant part of the deaths: some infected patients from elsewhere were put there under the "promise" of keeping them separate from the others, and you can imagine what happened later.
> Importantly the peak doesn't last long - epidemics don't spend 6 months with a stable number of people getting sick at the peak. Look at graphs of COVID deaths and cases and you'll see the standard pattern.
What's interesting is that many US states have shown virtually linear growth for more than a month, which is an extended "peak", assuming it's a peak. That depends on if enough of the population is exposed before the interventions are relaxed, or else it will just revert to its natural progression and the media will have something exciting to report on again.
> What's interesting is that many US states have shown virtually linear growth for more than a month,
The US data is tricky to interpret because the nation's testing rate has been growing relatively slowly. Over the past week (Covid tracking project data), the US has reported about 242k tests/day; for the week ending April 7 (so 4 weeks ago), that number was 144k tests/day.
In the meantime, an extended peak is consistent with the idea that policy measures in place have reduced the R0, but only to a value close to 1. Suppose stay-at-home orders reduce the number of daily contacts by about 70%, taking the R0 from 2.5 to about 0.8. With an infection period of two weeks, that would only reduce the number of new infections per day by 35% after a month.
> The UK, France, Spain, and Italy are all countries that have temporary morgues because of covid-19.
Aside due to the fact that there were high number of cases and the skewed death rate towards the elderly, couldn't also this be due to the fact that (at least in Italy) tests are done only if a patient is hospitalized?
I mean, the evidence is so far scant and anecdotal, but the timeline for treatment (I'm aware only remedisivir has been proven to be effective, but protocols also use other drugs, even if the efficacy is unknown) suggests that the earlier the treatment, the more effective it will be.
If only admitted patients are tested, that usually means a lag from symptom onset, which may ultimately be detrimental.
> Disease spread tends to always show a pretty standard shape. It was that observation (Farr's law) that led to epidemiology being born as a field. It grows, it peaks, it enters immediate decline.
Only without intervention. With social change, the disease's R0 (with respect to a given society) will differ, altering the shape of the epidemic curve.
We can have great confidence that COVID has not gone through a full, "status quo" epidemic curve anywhere.
Take Italy as an example. That nation has conducted (as of May 2, Wikipedia data) 2.11 million tests to find 210k positive cases. That 10% test positivity rate forms a loose upper bound on the prevalence of nCoV in the entire population -- even if we assume that policymakers erred and many infections are asymptomatic, test-selection criteria should not have caused a worse than average chance of detecting a positive case.
In the meantime, Italy documented 28,710 COVID-related deaths as of that date. If we again make the generous assumption that all COVID-related deaths were detected but the true population prevalence was about 10%, that would give the disease an 0.48% IFR. That's far too high for a rapidly-spreading disease to remain hidden for long.
Simultaneously, we can't say that the supposed 10% infection rate is sufficient for herd immunity. If 10% of the population is infected now when the virus was introduced at the end of December, the R0 of the disease must be well above 2 (with a generously short two-week period between infection and recovery -- shorter than many observations -- we've only had 8 generations.) Herd immunity would then require > 50% immunity in the population.
Instead, the much simpler conclusion is that policy responses have worked, with lockdowns and distancing reducing Italy's R number to somewhere around 0.75 (based on a rough look at the number of new cases per day, divided by the number from two weeks ago).
> All we can do is work to piece together a narrative of what really happened by examining all the evidence we can get.
Yes, but we must examine all that evidence in light of what we know of epidemiology. It's far too easy to cherry-pick data that is comforting or aligns with our political predispositions.
> Repeated anecdotes from many different people about having had a COVID-like illness before it was being discussed much are interesting for that reason.
... but the plural of anecdote is not data. Especially for a disease such as COVID, where the range of attributed symptoms is so wide that just about any commonly-circulating cold or flu could -- by symptoms alone -- be attributed to nCoV.
Also, Lombardia and neighbouring regions are still by by far the worst affected. Southern regions are still relatively unscathed. While it is likely that there might be other factors that contributed to the difference, it is at least plausible that lock down was a contributor.
Italy will slowly get out of lock down soon (while still preventing travel across regions). We will see how it goes.
A lot what you said is just factually incorrect as pointed out by others. Those anecdotes will be interesting when they can point out any differentiation from regular seasonal flu (such as bilateral pneumonia etc)
