What's Goin' On?
Stanford's John Ioannidis, who really knows his stuff, has released a preprint of an article that analyses mortality rates of Covid-19 in folks > 80 years of age versus folks < 65 years of age. I found the preprint because it was referenced from a site that BlueRepublic had linked to in his COVID-19 Deep Dive - Swiss Propaganda Research essay. Here's what that website said in their item listing:
Stanford professor of medicine John Ioannidis concludes in a new study that the risk of death from Covid19 for people under 65 years of age, even in global „hotspots“, is equivalent to the risk of a fatal car accident for daily commuters driving between 9 and 400 miles.
As an exercise, it would probably be useful at this point for the reader to take a closer look at the SWPRS quote, and ask, "What exactly do I think Ioannidis and his colleagues found? What exactly do I think they reported as their result?" No, don't go off and do a bunch of research, just take a moment to synthesize.
The purpose of this essay is not to spark a debate with BlueRepublic, nor to attack that Swiss Propaganda Research website -- which if nothing else, looks like a good place to track down some of the sources for a lot of what goes flying around in the infosphere without accompanying citations and sources, so I thank BR for providing the link, even though I disagree with the gist of much of what is presented there.
My intent is instead to address the problem of how important information gets "digested" as it gets passed around on the tubes, and how, as with the classic game telephone tag, the eventual result can be confusion, or worse, nonsense, or even worse, straight-up misinformation masquerading as True Facts.
In this example, the SWPRS digest is "mostly" accurate, particularly with respect to the paper's abstract. Here is an excerpt from Ioannidis's paper, including the tag end of his abstract, as well as his stated conclusion:
... The COVID-19 death risk in people <65 years old during the period of fatalities from the epidemic was equivalent to the death risk from driving between 9 miles per day(Germany) and 415 miles per day(New York City). People <65 years old and not having any underlying predisposing conditions accounted for only 0.3%, 0.7%, and 1.8% of all COVID-19 deaths in Netherlands, Italy, and New York City.
CONCLUSIONS:
People <65 years old have very small risks of COVID-19 death even in the hotbeds of the pandemic and deaths for people <65 years without underlying predisposing conditions are remarkably uncommon. Strategies focusing specifically on protecting high-risk elderly individuals should be considered in managing the pandemic.
It's important to understand that an abstract is just that: It's a brief summary that dispenses with details (i.e. precision) in order to give a general overview of the paper's subject. It is always dangerous to base an opinion on the content of an abstract, and it is even more dangerous to pass along what is effectively an abstract of an abstract, because that missing precision can quickly transmogrify into inaccuracy.
The reason I think this might be useful is quite subjective: My own first impression of the findings, based on my own reading of the SWPRS digest, turned out to be quite far off the mark. And much of that mistaken impression was "my fault" -- I formed that first impression before I had stopped to think carefully about what the words mean. Which is why it was so important that, before I started blabbering about it to other people, I followed the link and read the goddamned paper.
Important preface: Like too many people, I have been regrettably casual in my use of the terms "mortality", "mortality rate" and "lethality" almost interchangeably. These terms do not mean the same thing, at least not in the technical use of epidemiologists and etiologists; Yet almost everywhere, "us folks" are using "mortality" or "mortality rate" when we should be using "lethality", etc. I am going to try to avoid that mistake here.
Okay. So. Here is my attempt to state in my own very simple terms, the part of Ioannidis's result that is described in the SWPRS digest:
These two likelihoods are about the same:
1. A person dying from Covid-19 as of April 4, if the person were
a. a resident of New York City on the day of the first reported NYC case
b. under 65 years old.
and
2. A person dying in a traffic accident somewhere in New York State, if the person were
a. Random person (no assumptions about age, residency, or anything else)
b. Driving 415 miles per day in New York State, over a comparable duration during normal times (ie., it's not a direct comparison to the current situation with greatly reduced traffic)
For the German case, just replace "New York City" with "Germany", "New York State" with "Germany", and "415" with "9". And by the way, even this attempt to nail down the details comes up short. For example, Ioannidis's stats are based on New York State's reported "road deaths"/"billion miles driven". I haven't hunted them down, but I'm pretty sure they include, for example, pedestrians who are run down by cars. Which is more people than you might want to know, especially in New York. In other words, that "miles driven per day" figure should likely be even higher. But I'm not sure about that, I'd have to go back to the stats source cited by Ioannidis to be certain.
Anyway: The most important thing to realize, before you share the Ioannidis result with other people, is that the comparison with driving certain distances refers to overall Covid-19 mortality rates for the population across the specific time period. It does not refer to the lethality of the disease. He is not talking about the chance of someone under 65 who has been diagnosed with Covid-19 dying of Covid-19, he is talking about the observed overall fraction of New York City residents under 65 who have been assigned Covid-19 as their cause of death since the onset of the pandemic.
So: I can't speak for you, the reader. Do you think you initially came away from the SWPRS digest with an accurate understanding of the paper's analysis and stated results? Do you think you were [EDIT to fix copy paste error] adequalty adequately prepared to discuss the issue with anybody else? (BTW, when I said it was "mostly" accurate, the one clear inaccuracy was the use of the word "commuter". None of the paper's analysis says anything about the "type" of driving -- those are fully abstracted miles driven, encompassing all the driving that everybody in New York State drives.)
Keeping that in mind, here are some other things to understand about the report and the numbers in it.
A. Ioannidis discusses various other facts and stats. The whole "miles driven" thing is not the meat of this report, it's the steak sauce.
B. Nonetheless, Ioannidis found that under-65s in general had low Covid-19 mortality rates, relative to over-80s.
C. In the US, the fraction of fatal Covid-19 cases that were under 65 is much, much higher than in Europe.
D. Because Ioannidis's numbers include the time period when outbreaks were just ramping up in the hotspots, they may underrepresent the "true" mortality rates. His rates would be a lot higher if he'd done the analysis on the three weeks from March 27 through April 16.
E., Because Ioannidis's numbers end at what might represent a peak for the outbreaks, they may overrepresent the "true" mortality rates. We don't know, because we don't know what the future looks like. Maybe the numbers will crater over the next 3 weeks.
F. Notwithstanding the recommendation of his conclusion, Ioannidis does not directly analyse or address the extent to which "lockdown" policies might be disproportionately suppressing the prevalence of Covid-19 amongst younger people, and thus the extent to which his observed result may depend on such policies, notwithstanding the recommendation of his conclusion.
G. Ioannidis's conclusions explicitly do not address the question of non-lethal consequences of the infection, in old folks or in anyone else.
H. Etc.
BTW, driving 415 miles per day is not actually a low-risk proposition, as such things go.
Comments
GIGO. The data being compared
...as I believe you point out — are disappointing at an elementary level. It's an article I would skip in the interest of time. Although I might look at it just to confirm my instincts. For many, however, the data groups being compared look very close and promising — and in a way, that's a good thing. I would encourage interested parties to follow all such sensational "statistics" to their natural end — because by the time this epidemic is over, those who study these intellectual failures will become the wise among us, able to spot garbage-in-garbage-out instantly.
We need more experienced skeptics (with on-the-job training) in the United States of Propaganda — in preparation for what lies ahead.
** As a footnote, let me point out that I do believe that there are certain analyses about data out there that are currently described as CT, and those who mention it are vigorously punished by self-appointed authorities. I am not writing about this information myself, but I can see no verifiable evidence that suggests that it should be rejected. Until opposing physical evidence comes into existence, this information remains viable in my mind. This strenuous rejection makes it more interesting, if nothing else. Do a better job, Thought Police!
____________________
The political system is what it is because the People are who they are. — Plato
I read the paper for two reasons:
A. I've been following Ioannidis's work for over 10 years. I'm confident that he knows what he's doing, as far as analysis goes.
B. I'm certain this paper will be cited all over the place over the next few days, and I'll be very interested to see the extent to which it is misquoted and/or misunderstood.
That's why my focus was not on the results themselves, but only on the differences observed amongst:
A. The digest of the results at SWPRS
B. The abstract of the results from the paper
C. The full detail of the results in the paper
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
Yeah, this is confusing.
I think they're relying on general actuarial tables and then extrapolating based on the mortality/lethality/death/whatever rate of the locality.
So basically the COVID death rate for Germany is equivalent to the odds of a fatal accident if you drive 9 miles a day (very low) while the higher death rate for NYC is equivalent to the odds of a fatal accident if you drive 415 miles per day (higher).
I actually took this line yet another way when I first saw it: I thought they were saying the average death rate was equivalent to driving in Germany 9 miles a day.
Then I thought. Wow! That Autobahn is really dangerous!
The current working assumption appears to be that our Shroedinger's Cat system is still alive. But what if we all suspect it's not, and the real problem is we just can't bring ourselves to open the box?
I, too, had the very immediate reaction of,
Jesus Christ, are the Germans completely unhinged on the road!?!?!?!
Okay, not really, what I really thought was, "Uh... don't think I'll be driving in Germany. Ever."
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
Replying Late but..
Just wanted to say that I interpreted this the same way as you did at first. 'My Goodness, driving in Germany is dangerous!' Then I rethought. Heh.
If you're poor now, my friend, then you'll stay poor.
These days, only the rich get given more. -- Martial book 5:81, c. AD 100 or so
Nothing ever changes -- Sima, c. AD 2020 or so
A few thoughts
First I did not see a link when I went to BlueRepublic's diary so I could not analyze the paper. There has been much BS associated with this outbreak, however. Second, people under 65 may recover, but some recover only after hospitalization and they are sick as hell. Look at Boris Johnson, for example, who says he has not doubt that he would have died it not for the NHS. Relatedly, if we don't limit the spread of the virus as much as possible, hospitals will be full and, as in Italy, doctors will have to decide who lives and who dies. Third, we already have 26+ thousand deaths in the US. If under 65's who have no underlying conditions are allowed to spread the virus freely, they will certainly infect millions who will have a high death rate.
RB -- sorry about the missing link.
BR's essay links, right down at the bottom to this place
You can get a PDF of the paper itself right here
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
Oh, and I agree with your other remarks,
some of which I touched on in the OP.
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
Second and third waves need to be feared
We've only just begun w/wave one the least harmful
wait for the next 2 waves will be the real action occurs -nfg-
history.com
Reported cases of Spanish flu dropped off over the summer of 1918, and there was hope at the beginning of August that the virus had run its course. In retrospect, it was only the calm before the storm. Somewhere in Europe, a mutated strain of the Spanish flu virus had emerged that had the power to kill a perfectly healthy young man or woman within 24 hours of showing the first signs of infection. In late August 1918, military ships departed the English port city of Plymouth carrying troops unknowingly infected with this new, far deadlier strain of Spanish flu. As these ships arrived in cities like Brest in France, Boston in the United States and Freetown in west Africa, the second wave of the global pandemic began.
“The rapid movement of soldiers around the globe was a major spreader of the disease,” says James Harris, a historian at Ohio State University who studies both infectious disease and World War I. “The entire military industrial complex of moving lots of men and material in crowded conditions was certainly a huge contributing factor in the ways the pandemic spread.”
I never knew that the term "Never Again" only pertained to
those born Jewish
"Antisemite used to be someone who didn't like Jews
now it's someone who Jews don't like"
Heard from Margaret Kimberley
The US currently has 4 aircraft carriers out of commission
Yanked out of the oceans, they were.
There's nothing like 5000 people hot-bunking to give you a real viral experience.
The US has been broadcasting warnings to the world every hour saying, "If you try to hit us when we are down, we will wipe you off the map!" Or, words to that effect. The world is going, "Whatever, dude."
Some lesser participants in the battle fleets are also being disinfected.
On another note, I was told that about 1,000 people must remain on board a nuclear-powered aircraft carrier at all times, to prevent disaster. So, there's that.
So, why are the 2nd and 3rd waves much scarier. They probably are, but what's the thinking that leads to that conclusion, I wonder?
[edit = typo]
____________________
The political system is what it is because the People are who they are. — Plato
I guess just plain human nature?
remember at 20-25 years old we were all indestructible.
Plus look around the good ol usa didn't lock down let alone close down.
I never knew that the term "Never Again" only pertained to
those born Jewish
"Antisemite used to be someone who didn't like Jews
now it's someone who Jews don't like"
Heard from Margaret Kimberley
I don't even read things that compare totally unrelated
death rates, there are so many ways to put one's finger on the scales. Comparing old to young populations, overloaded hospitals to ones that aren't, healthy populations to polluted ones, undiagnosed deaths, driving fatality rates themselves vary greatly. Here, in my state, getting tested positive for people 50 to 60 means 5% will die, and by far most positive tests are in the 20 to 50 age group though half the fatalities are over 80. Time enough to compare it to this or that after we stop people from dying. Our doubling rate is down, almost out to 2 weeks, but we're still doubling.
In the US yesterday we lost like 20 less to covid than to heart disease, it's almost the #1 cause of death in America, and quite a few nurses are dying in their 30s. Until young docs and nurses stop dying needlessly I'll do whatever I can to help them not die trying to help me.
The list of what we don't know is long, yet people
everywhere are adamantly making very firm assertions about what is really happening. This would be bad enough even if most of those people would be unable to figure out was really happening even if they did have perfect information.
Some things we don't really know:
The clinical sensitivity of the now several various real-time RT-PCR test kits.
The clinical specificity of the now several various real-time RT-PCR test kits.
The clinical sensitivity of the now several various serum-based antibody test kits.
The clinical specificity of the now several various serum-based antibody test kits.
The clinical presentation the previously-known beta-coronaviruses already believed to be endemic in the human population.
The prevalence, regionally or globally, of the previously-known beta-coronaviruses already believed to be endemic in the human population.
The prevalence (or even existence), regionally or globally, and/or clinical presentation, of any previously UNknown ancestor of SARS-Cov-2 that might have already been in circulation before whatever mutation happened that made it the bad-ass em-effer that it is.
The relative "effectiveness" of the various transmission pathways (e.g., airborn droplets) of SARS-Cov-2.
The etiology/pathology of the damned thing. We don't know exactly what it's doing, we don't know why it's so deadly to some people and apparently mild in others.
The clinical effectiveness of potential treatments.
* So far, we're getting conflicting information from all directions. Does hydroxychloroquine help? Some small-scale studies indicate no. Other very-small scale "studies" suggest ... yes, maybe?
* Are NSAIDS a really bad idea? Or a really good idea?
* remdesivir?
Are China's numbers for real? How?
If everyone on earth were infected with SARS-Cov-2, and had suitable medical treatment available as is currently best understood,
* How many would be effectively incapacitated (including, just bedridden at home)?
* ... for how long?
* How many would require hospitalization?
* How many would die?
* How many would clear the virus, but have serious short-term complications?
* ... long-term complications?
How resistant are survivors to future infections?
Will this virus mutate at rates and in ways similar to influenza, requiring development of annual vaccines that will have only partial effectiveness as multiple strains bubble around in the population?
And so on.
Yet, despite the fact that none of these things are known, a bajillion blatherers all over the world are blathering out their firmly-held opinions, even though their opinions might literally cause the specific deaths of specific individuals who encounter those opinions, and even though most of said blatherers are insufficiently competent in the related sciences to be able to formulate a well-informed opinion even if all of these things were known.
I'm losing my patience.
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
And, some other things we don't know:
What is the mutation rate in Covid-19? And, is it possible that, as I think Roy alluded to above, that a covid-19 mutant could arise that readily kills healthy young folks? (i.e. similar to the 1918 flu pandemic). I think the answer is: We just don't know. It is too new.
However, it does strike me that, given the possibility of a mutant that could be much more deadly, a good way to keep that from happening is to keep replication of the virus and exposure of people to a bare minimum. It could well be way too late for that if only a few mutations result in a deadlier virus. But if it takes many mutations to get there, it might still be best to minimize new cases and hope it burns out.
Welcome to science. If we always knew what we were doing, it wouldn't be science at all...
To me the most important thing we don't know...
Is how many people have already had the virus and were asymptomatic? We need wide spread antibody testing.
The other looming question IMO is re-infection.
“Until justice rolls down like water and righteousness like a mighty stream.”
my suspicion is that people who are hoping that
covid-19 is normally so mild that most cases go undetected are going to be disappointed. but that is only my own intuition -- a guess, really -- based on my own survey of what data is available.
meanwhile, after i posted that link about remdesivir, new anecdotal news (from an ongoing clinical trial) came out suggesting it might be the real deal, and gilead's stock bounced (and with it, the entire market). gilead folks cautioned that we won't know for the full results for a while yet. if it's really effective -- if it can knock the symptoms down in just a couple of days, then maybe we're out of the woods.
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.
That's my guess as well.
In the few places where contact tracing was initiated early on at least a few of these postulated early mild and recovered cases should have popped up but so far they haven't.
Reports of re-infection spooked me.
It sounded like they were in people with mild symptoms, but if these reports are correct what about vaccines?
Also, will we have to worry about antibody-dependent enhancement making the second infections worse and creating problems for vaccine development? Immunology is not really my thing, but this is scary.
Until KCDC has investigated and reported
on these recovered and re-tested positive cases. there's not much to go on. If they followed SOP, recovery should only have been assigned after two negative tests. However, as with all tests there is a false negative factor. So, was SOP followed in these cases, were the tests properly done, and/or does the test produce some false negative results.
The age related re-tested positive suggests a few things. 75% of these cases are under the age of 60. Suggests that they were mild cases which in turn suggests that they recovered at home. More quickly than most? Young and healthy enough that the administration of the negative tests was somewhat slip-shod?
My (wild-ass) guess is that they hadn't fully recovered. This thing doesn't kill quickly and recovery is a longer process than annual influenza. They appeared to be symptom free and the viral load was too small to be detected by the test. Then once up and about, the virus did its thing again.
Gonna get yourself on a list
(as if everyone Isn't already) if you go around promoting dangerous stuff like critical thinking.
(don't forget apply it to climate science, too)
Thanks for the essay. Your "telephone game" analogy is apt.
I remember as a classic example of how statistics could mislead as the old Volvo commercial that glowingly asserted that (IIRC) "90% of all Volvo's sold in the last ten years are still on the road."
Which sounded good, except that you couldn't really determine that without knowing knowing the distribution of sales over the ten year period - if 90% of sales had been in the last two years it could actually be pretty *bad*.
Basic statistics was challenging enough for me that I try and tread carefully dealing with them - means and averages can conceal as much as they reveal, definitions and assumptions matter.
For example, while the average risk to a random driver driving that 415 miles in NY is quantifiable and no doubt useful for something, its usefulness in predicting the risk the risk to any particular driver in a real world situation is limited since there are any number of factors - sobriety/experience/emotional state/alertness etc. of the driver, season/weather/time of day, nature of roads travelled, condition of and driver's familiarity with the vehicle, etc. that would influence the actual risk. A useful distinction to keep in mind when making real world decisions.
With your post in mind I checked on the research behind the article referenced this way on the Swiss Propaganda Research site:
which linked to the Guardian article headlined Air pollution in China is killing 4,000 people every day, a new study finds
Physicists at the University of California have found 1.6 million people in China die each year from heart, lung and stroke problems because of polluted air
Which included bits from an interview with the lead researcher:
"To put Chinese air pollution in perspective, the most recent American Lung Association data shows that Madera, California, has the highest annual average for small particles in the United States. But 99.9% of the eastern half of China has a higher annual average for small particle haze than Madera, Rohde said.
“In other words nearly everyone in China experiences air that is worse for particulates than the worst air in the US,” Rohde said.
In a 2010 document the EPA estimated between 63,000 and 88,000 people died in the US from air pollution. Other estimates ranged from 35,000 to 200,000.
Unlike the US air pollution in China is worst in the winter because of burning of coal to heat homes and weather conditions that keeps dirty air closer to the ground, Rohde said."
(Note that the Wuhan outbreak occurred in winter and that Wuhan is situated in a basin subject to inversion conditions)
Using the author and journal's name I was able to locate the article itself
Air Pollution in China: Mapping of Concentrations and Sources
Robert A. Rohde ,Richard A. Muller
Which was using data from 1500 of China's own reporting stations over a four month (April to August) period - something apparently not possible before 2012 when China set up an accessible
network.
There were some differences between the Guardian headline:
"Air pollution in China is killing 4,000 people every day, a new study finds"
and the abstract:
"The observed air pollution is calculated to contribute to 1.6 million deaths/year in China [0.7–2.2 million deaths/year at 95% confidence], roughly 17% of all deaths in China."
Potentially big difference there, but it seems to resolve more toward the "killing" conclusion in the discussion section:
"The conversion of pollution concentrations to mortality is complicated. We adopt the framework [20] developed for the WHO Global Burden of Disease study [21] that considers PM2.5 mortality due to impacts on five distinct diseases and accounts for nonlinearities as a function of concentration. Using prefecture level population and pollution data along with national average death rates for the five modeled diseases, we calculate that 1.6 million deaths / year can be attributed to PM2.5 air pollution under the WHO model [95% confidence: 0.7 to 2.2 million deaths/year]. This is equivalent to 4 thousand deaths / day or 17% of all deaths in China."
But, that's assuming that the WHO Burden of Disease framework is valid...
Interesting jaunt.
Using the above numbers, it looks like China's normal daily death toll is about 24,000 so a little under nine million/year. If the official Chinese COVID-19 death toll is also accurate (a bit over 5000 even with the figures for Wuhan bumped 50% today) then the death toll would have had to be something like seventeen times higher to bump the Chinese annual deaths by even one percent.
Considering they shut down the bulk of their economy for a relatively tiny result, kind of raises the question of why would they (literal and ideological heirs of the greatest mass murderers in history) do that instead of just taking the hit and the economy going? Moral qualms seem unlikely somehow.
i can think of a few speculative reasons.
not all of them reflect well on humanity in general.
in any case, your example is a good one. the guardian headline says "is killing", the abstract says "contributes to," and the main text of the article elaborates on some of the difficulties of trying to assign blame for morbidity or mortality to one cause or another.
The earth is a multibillion-year-old sphere.
The Nazis killed millions of Jews.
On 9/11/01 a Boeing 757 (AA77) flew into the Pentagon.
AGCC is happening.
If you cannot accept these facts, I cannot fake an interest in any of your opinions.