Much appreciated and no less critical Virusvaria reader Jan van der Zanden wanted to see the 5-year cohorts per 100K. The Norm Mortality raised questions, which we will answer later in the article. We present the standard mortality as a baseline to keep it as clear and simple as possible, with the necessary connection to the graphs of our government and institutions. But indeed: details are lost with every merger. Those details can be interesting too. This of course also happens when 5-year cohorts are merged, to a certain extent, but much less than with cohorts such as 0-64, 65-79, 80+. The fact that more can be seen in 5-year cohorts is evident from these graphs. A category 0-30 or 0-40 wouldn't be out of place either, but see for yourself.
With the graphs:
- In each graph, the right part is tinted, starting with 2021. 2020 is therefore always the last year in the white part.
- The first open fold-out contains 21 graphs starting from 2010, including a fold-out from 1980 for those who want to look further back.
- Each cohort has its own graph. Combined graphs with more cohort lines are more difficult to compare due to the scale differences in the figures.
- The minimum and maximum values ​​of the Y-axis are now always aligned with the lowest and highest values ​​shown in the graph.
- Bear in mind with these graphs that there are much fewer deaths in the younger regions than in the older ones. This also means: larger differences, the occasional strange peak (or trough) and much less influence on the total excess mortality. When we total, we take the size of the age groups into account. This is done with 1-year 'cohorts', which are composed of weekly measurements that are combined into monthly measurements.
- Because Jan expressly did not want to see any 'modelling', we did not add any trend lines or baselines. After all, that is the only modeling in the Mortality Standard: the baseline used when calculating excess mortality.
Assignment: imaginary baselines
The assignment is: think of a trend line from a year of your choice, between 1980 and 2019 (this can also be a curve (this is even mandatory for a longer reference period), and then extend it in your mind into the tinted area.
Have fun again.
2010-2024 – Mortality NL: Graphs per 100K, 5-year cohorts
1980-2024 – Graphs per 100K, 5-year cohorts
2019 as a reference year – or not?
At the same time you see a large, global wave movement in the green band. We pass a green marker over it. Then we zoom in on 1950-2024.
We're going to work the KISS way for a while. That is to say: only with the 'Crude' mortality per 100K observations, and then extend linear trend lines with different reference periods. We will then explain why this does not give a correct picture.
In the graph below, the objection of “2019 as the last reference year†(Bonne!) has also been taken to heart. This persistent objection means that we would calculate excess mortality too high, because flu-free 2019 is the last reference year. This lowers expectations slightly. (You could also leave out 2018 because it was a year with a bad flu, but whatever.) The graph “trend lines up to and including 2018†contains four trend lines, all four with 2018 as the last reference year.
Only with the trend line from 2010-2018 will we get towards 2024.
We also see here, despite all trend line attempts, that a mortality increase such as that in 2020 and beyond has never been seen in the last 70 years.
What exactly does Norm Mortality do?
The population structure has a major influence on the total mortality per 100K. 'More people in their octogenarians' means for the total population: 'more mortality per 100K'. The extension of trend lines based on 2010-2019 (dashed lines) is based on the assumption that developments in population structure and mortality rates will continue. And while the population structure for 2020-2024 is already known! Using the actual population structure obviously gives a more realistic picture of the situation.
It can also be calculated what this structure will look like in the years after 2024. Roughly speaking: today's people in their sixties will be those in their seventies in 10 years' time minus their mortality risk in those 10 years. What we do not yet know is the number of births and migration, so these always remain uncertain factors in all forecasts. The most realistic expectation is shown by the blue line: the Standard Mortality Rate.
This was another job well done. So let us also say: help virusvaria.nl, steig.nl and redactiemonitor.nl keep the sites live and alive. It takes time.
References
- 1Sounds crazy? ChatGPT explains out here
Although I appreciate that you also reflect on other views, I think this is a short-cut explanation. Mainly based on crude mortality per 100k, over a 'very' long period of time.
I would not be so quick to include the 50 years in a forecast for 2025.
But in my opinion there are much more important side drawings to be made for 'norm mortality', which is not much more than adding up all 1-year male/female cohorts together.
And why 2018 is a much smaller outlier compared to 2019.
You could confirm or refute Bonne Klok's point by elaborating on five-year cohorts. Unfortunately that was not done. The yellow and gray lines could then run differently compared to each other.
How then, Hans? Can you show that?
I can certainly show it - but I can't paste any pictures here. Then a link to Twitter: https://x.com/HansV_16/status/1981346520050860288
(The tweet in question has been added in the reply below.)
Hans, I have added the tweet in question here. You write on X:
"Deaths in NL per 100,000 people, composed of five-year groups, total M/F. Expectation based on regression with starting year 2010 and ending with 2018 or 2019. Why this difference? 2019 was a year with lower mortality, which lowers the expectation for later years.
But that effect appears to be limited, even without 2019 there will always be excess mortality after 2019.â€
The crux of the whole piece is of course that if you put all observations per group of five years in a row, your brain will already see a trend break in many groups.
Do all points in the groups lie on an imaginary straight line? No, not after 2019.
Or on a curved line? No, neither.
Your brain tests the observations in a fraction of a second, without you even realizing it, against a very simple model for expected mortality.
That's exactly right. And if you see that mortality has returned to 2019 levels, or even slightly below that for a number of groups, then the conclusion is that there is no longer excess mortality.
Unless you have unshakable confidence in medical science, healthcare and people's lifestyle that their mortality will continue to decline rapidly.
I don't have that confidence at all.
That 2019 level has not yet been reached among men aged 45 – 54 years. So excess mortality.
This is (barely) the case with the rest: so no more excess mortality. There was in 2020 (which is still understandable and OK), but especially in 2021 and 2022 in the older categories, which is not OK.
At 90+ the trend was already increasing, I suspect due to an aging population, i.e. a gradually increasing average age in the cohort. If you were to split that cohort into 5 x 1-year cohorts, I suspect that the deaths would look "neater".
Jan, if you look here:
https://hansverwaart1.substack.com/p/de-verwachte-sterfte-bij-hart-en
Then you will see that it will become very clear that no structural improvement in mortality from cardiovascular diseases occurs or is possible at a very old age. Especially in men, you see a lot of improvement with high R2 values ​​until old age between 2010-2019, so a strong correlation.
You actually do the same thing as Anton, but you do it more transparently, because you calculate the mortality per 5-year cohort and only then add it up. Fine.
But some comments:
In the graph with 65-70 you can see that mortality is already well below that of 2019. But because you only accept straight lines (linear regression) in your model, this still results in significant excess mortality.
But: statistics are not always there to model reality for the future. But primarily to describe it. I just object to that linear regression for the future. Because: you simply cannot irrefutably maintain that mortality from 2020 to, for example, 2030 should decline at the same pace as from 2010 - 2019 or, if necessary, slightly less sharply from 2015 - 2019. That is the crux. And you, like Anton, apparently assume that
With Anton, he also lumps everything together with the concept of life expectancy. Then it all becomes very opaque...
Reality does not follow your (or Anton's) statistical model [however mathematically correctly described with the R-values], but reality follows what happens in reality with regard to lifestyle, care, etc. We do not seem to agree on this.
I therefore seriously object to this type of reasoning: "The 10-year trend line has a high R2, the 5-year trend line is much lower, so it seems that the choice to use the 10-year series as an expectation works out well here. The values of 2015 and 2019 mean that a convex trend is also possible. This would mean that in the second part of the period considered, the blue line is roughly followed, causing excess mortality in 2022 and 2023 with resp. half and 65% are reduced.â€
Mortality in the future does not follow any model of your “best†and does not have to do so at all. So R1 is not better or worse than R2. It simply follows what is present in the “real†world in terms of health and care. And if that is not better than in 2019, then the curve will simply flatten (completely).
I repeat: of course it is somewhat more likely that there would still be some decline after 2019 if “nothing had happened†then. But it is not a dogma. And so CBS/RIVM can make this statement based on evidence.
What they cannot say, and you rightly point out, is that there was no excess mortality in 2021 - 2023. Then there was no more Corona and the mortality per 100,000 was clearly above the level of 2019. Already from the category of men aged 50+. So that was really wrong! So I think that's where the focus should be. Because I think we can win that battle. Those with those “standards†cannot be won on the basis of evidence.
Agree Jan! A model, however complex it may sometimes be, is a simplified representation of the much more complex reality.
The decline in mortality in the short term can be described quite well by linear regression, usually a descending straight line. Whatever can be used as a baseline.
But we also know that mortality cannot become 0, so that downward line must decrease less at some point and must therefore become convex.
In another blog on my substack I compared several curves for Cardiovascular Disease: https://hansverwaart1.substack.com/p/nogmaals-sterfte-hart-en-vaatziekten
I also included concave or convex curves there. Especially at higher ages we sometimes see (temporarily) concave curves.
In those graphs I have also projected the official CBS forecast for this cause of death. Fortunately, Anton has been a follower of these blogs for a long time, but I have noticed that a long blog on Twitter attracts more attention.
@Hans, OK, then we apparently agree.
Then we agree, Jan! A linear regression line is indeed a useful rule of thumb for the short term (several years). So you will have to end up approximately* on that line again. That has always happened and should happen now. It was also the assignment in the article: come up with virtual regression lines. Not: pick a point. A single point is not enough to draw a substantiated line, for which you need at least two points.
*The fact that that line bends slightly in the horizontal direction and that demographic wave movements disrupt it is a matter of fine tuning. Demographically, now that the baby boom is slowly disappearing, little special can be expected in the coming decades. But feel free to make a slight curve in your mind's eye in the regression line you see up to and including 2019. And if you're curious about how it fits into a long-term trend, that's why I made that second set of graphs.
For a detailed explanation I refer to ResearchGate. It is mainly about taking into account the shifting population structure and not blindly going along with increased mortality risks. With these long-term data, I wanted to prevent them from being pulled out of a hat like a rabbit. You see a clear wave driven by population structure. Important factor that should not be missing from forecasts.
You write "A linear regression line is indeed a useful rule of thumb for the short term (several years). So you will have to end up on that line approximately* again. That has always happened and should also happen now."
No, no, no.
You then do exactly what Ronald Meester fundamentally objects to in his explanation of his nitrogen research in DNW, in my opinion 100% rightly. And that is: declaring the easiest or most obvious or least complicated mathematical model applicable to “Realityâ€.
In principle, you can never predict the future with certainty using a linear regression line from the past. Unless you can clearly demonstrate that all relevant parameters will behave in the future as they have in the past. And you certainly can't do that in this case.
You can make it very plausible, but even that still cannot prove, that an ever-achieved lowest mortality/100,000/age category (generally in 2018 or 2019) should also be achieved again after a bad flu/Corona [in 2020]. And that is why I am setting the standard at that figure for the time being. Again that is conservative, but irrefutable.
You are a lot more optimistic about progress in lifestyle and healthcare; but you can count on it that you will find it extremely difficult to convince the world outside our bubble.
The key question then is: do you want to convince others? or do you want to continue to defend your own right about something speculative without external success?
I prefer to choose the former. I repeat: not because I fundamentally disagree with your speculation; because I also think it is quite likely that more decline after 2018/2019 is obvious. But I realize too well that I cannot possibly defend that scientifically. And by persisting in that speculative norm, my entire message is undermined.
Compliments for these basic data!
I took a closer look at the graphs from 2010 – 2024 in particular.
Then this confirms what I already got from Bonne's data:
1. Up to the age of 45 it goes in all directions and in my opinion you can hardly speak of excess mortality. But downward trends are generally stopped.
2. From the age of 50, you will see serious excess mortality, especially among men, from 2020 (!) and especially in 2021 + 2022. That's not OK!
3. From the age of 60, this effect will virtually disappear before 2022. But definitely not for 2020 and 2021. Here too, much stronger among men than among women.
4. What is striking is that only among 65 - 74 in 2021 (vaccination year) it is worse than 2020. This is not the case in the other cohorts.
5. There is an increasing trend at 90+ and 95+; I assume that this is an effect of an aging population, i.e. inhomogeneity in that 5-year period that shifts internally. In 1-year cohorts this would probably look different...
The conclusion is practically identical as in my previous post to the previous article:
A. From the age of 50 - 60 (and also somewhat in 45-50 years), excess mortality, especially among men, in 2021 and 2022.
B. Excess mortality from the age of 60 in 2020 and 2021, more for men than for women.
C. Excess mortality will have largely disappeared by 2024: mortality in almost all cohorts has returned to pre-2020 levels or even slightly lower. The downward trend is generally stopped completely. But well, can you still speak of excess mortality?
In short: there are hard signals, especially in 2021 and 2022, that things were seriously wrong. I hope it will emerge whether or not this is linked to vaccination.
From 2024 onwards, I do not find the figures that worrying anymore, even though the previous downward trend has stopped. But hey, let's make it clear that decline should be the norm...
In my opinion, the last “collective graph†provides a more concise picture of excess mortality in 2020, and in particular 2021 and beyond, than the underlying detailed graphs suggest. That's really crazy!! Is this the result of distortion due to a “model†used? Or just optically because of the different scale?
Mi. your excess mortality analysis should therefore not be based on these types of “cumulativesâ€, but on the collection of detailed graphs per 5-year cohort.
Even though excess mortality appears to have largely disappeared from 2024 onwards, given the clear alarm signals of 2021 and 2022, it remains interesting and relevant to investigate whether there is a relationship between vaccination and higher mortality. Even now.
You said it right: the excess mortality APPEARS to have largely disappeared. However, you should not only look at the output, but also at the input.
The situation is more worrying than you can initially imagine, at least if you take the changing population structure into account. It's now about twice as serious as it seems. We are indeed convinced that the decline should be the norm, see Researchgate. You can calculate it yourself, the files are included.
But apparently we also look at the 100K graphs differently. I also see an increase at 30+. And I do find it particularly worrying that downward trends among young people have stopped at the same time that excess mortality is occurring among the elderly. You don't just let that happen in this situation.
We have been well above a baseline based on demographics (not just mortality) for five (5!) years now and because it appears to have become less than it actually is, we are reassured...? 40,000 deaths later!?
Keulemans is right: As long as people don't drop dead en masse, we can simply continue with the order of the day.
They do not die en masse, but they do very regularly with the disease "suddenly", the sick famous Dutch people and athletes have also at least doubled (including young people) and/or their parents have become ill or died young. Changing to their order of the day is incomprehensible. Often a book is still written (disease gain). Declining birth rates and declining healthy births will also have an impact. You see the weakening of the effects of the c-shots everywhere. In a new media talk show, a very critical man with a lot of knowledge sat opposite a 19-year-old boy who has 3 injections in his body for which his parents are partly responsible. I suddenly hear that man downplay the effects of the injections and call the boy a healthy young man... From now on I will not downplay it anymore, I promise!
What do you mean by “also look at input�
If I see a pattern where mortality is even below the 2019 level, I find it difficult to maintain that there is still excess mortality.
Continuous (strong) decreasing trend in mortality is not a law of the Medes and Persians. Young people are starting to smoke/vape again, they exercise less, they are stressed due to housing shortages, etc. etc. All factors that do not exactly have a positive effect on health.
2021 and 2022 are really super suspicious. And 2023 in its slipstream.
For the rest, I don't think this can be determined from the graphs. This does not alter the fact that, with proper research, you could come to the conclusion that the undecreased mortality is mainly among vaccinated people. So I certainly advocate that research. But in my opinion, it is difficult to argue based on evidence purely from mortality that there is still excess mortality. C.q. It is well justified with the same evidence that excess mortality will have virtually disappeared from 2024 onwards. Again, looking at the detail charts.
Your “collection graph†suggests something very different indeed. But there is a model underneath that is open to discussion. And I really attach more value to the interpretation of the graphs per annual cohort, which have not been tampered with, but which purely show the mortality per cohort per year. I think that is “evidenceâ€.
That is why, in my opinion, further fighting for recognition of this type of excess mortality is rather hopeless.
Do you have a watertight argument why your collective graph is stronger evidence than the interpretation per 5-year cohort?
By “input†I mean that population structure does not play a role in those cohort graphs. If you just want to look at it this way, there is just as much excess mortality in a very aging country as in a country with a young population. But then you are interested in mortality rates, not in excess mortality. But mortality risks do not necessarily result in hospitalization, ICU occupancy, funerals and mourning; excess mortality, yes. So that is a problem.
The large (excess) mortality (fortunately) does not come from the young, so we may notice that vaping and such later.
The major wave movement since 1960 has less to do with health than with population structure. That is why continued decline is indeed not a law of the Medes and Persians, it is largely predictable. I'll show you that later. In any case, this wave movement is not due to viruses or decreased gym visits.
If we are only now below 2019 levels, we have been set back 5 years. That's intense. After all, we should have been there again soon after Covid. You also look at the downward trends before 2019, not just at the 2019 data point, right? That is a very shaky reference: taking the mortality of 6 years ago as a reference and if we are below that, it is OK...?!
By now we should be much further below 2019, combining all factors. Look at life expectancy, for example: if the recovery continues at this pace, we will not be back on schedule until around 2030. That's really not normal.
Proper forecasting (from 2020 onwards) is not possible without taking the changing population structure into account. If you do that, you really see that you should now expect less mortality. That's the receding blue line. Do the math, this apparently hand-waving is a bit too easy for me. But I also understand that you don't want to study all the pieces on RG.
The “2019 argument†is therefore not final evidence. In virtually every age group that plays a role in mortality rates, mortality per 100K is stagnating or growing, while it should be declining. I'll come back to it later.
I hope that the BMRK can one day force the release of correct deltavax files, then we will see that this story is correct.
(That's why they never get them, I could also say. So wasted energy. But I'm in favor of a two-track policy!)
And there is some room for improvement with that two-track policy! Badly needed! In the meantime, people like Marjolein van Egmond are allowed to make assumptions in the mainstream media. “coroa shot as a medicine†and then only say in the last sentences that the “results†could also be because the very sick people were counted among the unvaccinated (or had already died in those 100 days between shot and examination? And how many have (recurring cancer) gotten from those shots? Where have we seen this loose piece of science before... It's cringeworthy.
Anton, I think we are talking past each other about what we want to measure.
You apparently want to determine the total excess mortality and therefore apply a correction model for an aging population. That is relevant for social planning: how much hospitalization, ICU capacity, funeral capacity do you need? This is useful for budget and policy.
But that does not answer the question: will there still be excess mortality after Corona? Possibly as a result of vaccinations.
That question is about: do people within their age group die more often than normal? You measure this by looking at mortality per 100,000 per homogeneous age cohort.
Your sentence “If you only want to look at it this way, there is just as much excess mortality in a very aging country as in a country with a young population†makes no sense for cohort analysis. A 72-year-old in the Netherlands has the same risk of dying as a 72-year-old in a “young†country (with equal care and equal health). The number of 72-year-olds varies, but that is irrelevant to the question of whether there is excess mortality within that group or whether there is a deviation from a "norm".
Especially from the age of 85+, broad cohorts (85-90 years) are misleading due to unstable ratios over time: there are always more 85-year-olds than 89-year-olds within such a cohort, but due to better care and health, this distribution gradually becomes more homogeneous (the ratio of 85-year-olds vs. 89-year-olds becomes more equal). And due to an aging population, the absolute number in that cohort is increasing. Because the relative proportions of the numbers within the cohort change, the mortality for the entire cohort increases from 85 to 90 - while the mortality per individual year (85, 86, 87, etc.) can remain stable. That is why I advocate 1-year cohorts at an advanced age. Only then can you clearly see whether mortality deviates per specific age, without distortion due to shifting age distribution within the cohort.
The difference in concrete terms:
• Your approach: “We have more old people, so more deaths. Adjusted for this, there is still excess mortality.†→ Relevant for: hospital capacity, budgets, funeral industry.
• My approach: “Does a 75-year-old now have a higher risk of dying than a 75-year-old in 2019?†→ Relevant to: is there still a health problem, do people die too early?
For the second issue (and that is what the social debate is about) the cohort graphs are pure evidence. If mortality stabilizes or returns to around 2019 levels in almost all cohorts, then there is no longer an increased mortality risk per age group.
It is logical and predictable that there are more deaths in absolute terms due to an aging population, but that is not excess mortality - that is demography.
The core:
• 2021-2023: clear excess mortality, especially in the 50+ cohorts, and especially men → people died more often than expected for their age. So suspicious, possible vaccination side effects.
• 2024: return to ~2019 level in most cohorts → no longer increased mortality risk. But no further significant decline with the exception of a few cohorts.
• The fact that mortality/100,000 does not fall as fast as your model predicts is apparently due to an aging population; not due to excess mortality per age cohort. And the latter is the crux of the matter.
Your correction model adds a model layer that contains assumptions (continuous improvement in healthcare/health, (in)homogeneous effects of aging). Those assumptions are defensible, but not hard facts. Other assumptions are just as defensible and so it will be difficult to get recognition for this.
My question: Given my response above, what do you think is the strongest argument why your modeled baseline is better evidence than the direct cohort data for the question “is there still excess mortality?�
P.S. You really challenge me to think about this incredibly sharply and clearly... So thank you for your response.
That's mutual. It has also given me a new insight, I will write an article about that soon.
But there is no 'correction model' and I actually don't understand why you maintain that. The model follows the actual development of the population, per age, and calculates what you can expect in terms of mortality. It's that simple. You're just dealing with 104 ages and decades to plow through. That makes it a lot of work. It is not a matter of juggling derivatives.
The definition of excess mortality is simply: more than expected mortality.
Increasing life expectancy seems to have almost reached its limit, as you can see in the graph in another reply (this graph). It's an undeniably flattening curve. Is it really a questionable assumption that it will continue? How? If a kink suddenly appears there, you cannot point to an earlier year and say 'yes, but it was worse then'. Such a kink is a signal that something is seriously wrong.
About the health problem: with an age-discriminating disease, that problem disappears if you don't have elderly people (take Niger for example), even with equal mortality rates at every age. So it's not just about cohorts.
And a tip: if you take 2018 instead of 2019, there is even less excess mortality! 🤗 Why don't you do that?
Anton, thank you for your response.
I'm curious about your new insight...
You challenge me to think even more sharply, and that works both ways. I have some observations that I cannot reconcile with your story. Maybe you can explain that:
1. Life expectancy vs. cohort charts
You point to life expectancy as evidence that something is wrong. But that graph shows that life expectancy for men is back at 2019 levels, while that for women is lagging behind.
Your own cohort graphs show just the opposite: I see similar recovery in mortality/100,000 in most cohorts, or even slightly better than recovery in women in certain cohorts.
How do you explain that difference? Which of the two should I trust – the (indirect) aggregated life expectancy or the direct cohort data? Or did I not look properly?
2. Why 2019 would be problematic as a reference
You suggest that 2019 is weak as a reference (“Why don't you do that?†– referring to 2018).
Great point. I took 2019 because that was the lowest level in the series for almost all cohorts before 2020. And that was apparently already feasible with our care and lifestyle. So if that level is not reached, I think it is really an issue.
But then my question is: In what way would your cumulative model provide a better objective reference? You extrapolate a historical trend and call deviations from it “excess mortalityâ€. But why is that extrapolated trend more truthful than the actual situation of 2019 and slightly below?
4. The 'correction model'
You say: “But there is no 'correction model'â€.
Yet you also write: “The model follows the actual development of the population, per age, and calculates what you can expect in terms of mortality.â€
That's a model, right? You take historical trends, apply them to current demographics, and calculate what you expect. Cohort graphs don't calculate anything – they just show what happened.
What makes your calculation of 'expected mortality' more objective and better than the actual observation per cohort?
5. Niger vs. Netherlands – relevance?
You write: “About the health problem: with an age-discriminating disease, the problem disappears if you don't have elderly people (take Niger for example), even with equal mortality rates at every age.â€
But aren't we comparing the Netherlands with itself? The key question is: does a 75-year-old Dutch person have a higher mortality risk in 2024 than a 75-year-old in 2019?
And it doesn't matter at all whether Niger (or the Netherlands!) is much younger or older for that key question: if the mortality for a 75-year-old in Niger is the same as in the Netherlands, then apparently his health and care are just as good as in the Netherlands.
I fear that the average 75-year-old in Niger has a higher mortality - that is because the care there is worse and the lifestyle is probably also worse. Or maybe people who reach 75 there are much tougher and survive more. Anyway: that has nothing to do with the fact that there is an (on average) younger population there. That is not the problem we are trying to solve here.
We want to know what the life expectancy of a 75-year-old “reasonably†should be in our country. And if it is lower than in 2019, things are very wrong. If it is back to (approximately) the level of 2019, then things were bad in 2021 and 2022 (and for some groups also in 2023), but things have recovered.
The only dispute we can have among ourselves is whether it is logical that you can continue the downward trend, if necessary somewhat moderately, after 2019, or whether you put it on a trough/plateau. But, if all goes well, this has absolutely nothing to do with aging, life expectancy, etc. and therefore with your cumulative model of lumping everything together with the concept of life expectancy. These are all derivative concepts that also distract us from solving the issue we are facing.
6. The core question that remains
You have a model that says: “expected mortality must [continue] to decrease through continued improvement, so current level = excess mortality.â€
I see cohort charts that say, “mortality by age is back to 2019 levels.â€
If mortality is back to 2019 levels in almost all cohorts, how much can you say that there is substantial excess mortality? Are you actually saying: “the problem is not that people die more often than before, but that they do not die less often than I expected based on increasingly better care and lifestyle�
I hope you understand that these are not rhetorical questions – I'm really trying to understand why your cumulative approach based on life expectancy would trump interpreting the raw cohort data based on mortality/100,000.
This is an interesting discussion. What I miss, however, is the 'pull forward' effect. Attention has been paid to this in the past on this website via 'undermortality after excess mortality', and I think we cannot avoid taking this into account here. And then unfortunately modeling inevitably comes into play.
If hypothetically the vaccines have caused significant excess mortality in 25-year-olds, then you will see a recovery to the normal line as soon as the flu stops with boosters. But at 85+ you will actually see a dip in under-mortality, because some people died prematurely, but life expectancy in that group is such that 'only' a year or two is often lost. However, with a 45 year old you have to model because you have to assume whether someone has died from, for example, acute heart or pericardium inflammation (a recognized side effect of the vaccines), in which case no under-mortality is to be expected. Or do you assume that the excess mortality was caused by perhaps cumulative heart damage from such an incident (any heart or pericardium inflammation, no matter how mild, causes permanent heart damage that has a cumulative effect). In the latter case, you will mainly see 45-year-olds affected, who already had a weaker heart and therefore expect an under-mortality for a few years. You will therefore have to make cautious assumptions per age category - if necessary with healthy reluctance. Perhaps not when drawing up the graphs, but certainly when analyzing the graph.
In that context, an almost return to the pre-2019 standard line is not necessarily a good sign, because in certain categories you should be below that standard due to under-mortality.
What will cause the mortality peak in 2022 for 1-4 year olds (M and F)? Are they mainly 1-year-olds? Are they mainly born from vaccinated parents?
It also peaks in 2012 and 2016. So in my opinion it is difficult to define it as "worrying".
The data to determine whether parents have been primarily vaccinated is almost certainly non-existent. Then as a researcher you really have to dig into the details to first track down the parents of the toddlers and then check their vaccination status (which is also not easily available). So that will be very difficult to investigate...
It is also difficult with these small numbers. Yet 2022 is different. Look at the beautiful synchronization of M and F. That is quite strange and different from 2012 or 2016. Whether or not it is due to parental vaccination.
No, you also see this happening in 2012 and 2018/2019. And. It's a very random pattern. More intense in boys than in girls.
The absolute numbers are approximately x 1.6 (approx. 150,000 – 170,00 babies are born per year). So it's not very small either.
I don't really see a trend break in/after 2020.
The trend break is really evident at higher ages; especially considering the much higher numbers...
Could indeed be worse, see US. They fall there en masse. (Life expectancy also corrects for aging). Kennedy is still being ridiculed here. USA is not Country (anymore) for old men (movie is recommended, by the way)
https://share.google/images/dzbRqqOQcGzfMAYS5
The fact that the Netherlands does not recognize long-term excess mortality may also have to do with the definition of excess mortality:
https://www.cbs.nl/nl-nl/faq/corona/medisch/vragen-en-antwoorden-over-de-sterftecijfers
Excess mortality is by definition temporary. Doesn't seem to be long lasting. (Familiar short-term view?)
The question is whether a government, such as the US, should only take action if life expectancy drops sharply and not if it stabilizes or falls contrary to expectations.
When the suspicion of a permanent increase turned out to be incorrect, people were quick to respond (increasing the state pension age).
What I miss a bit is a broader view. I think it's excellent to argue from the Netherlands, but of course this discussion takes place everywhere. And I sometimes miss that element a bit. I mean that 2019 may have been a mild flu year for the Netherlands and significant aging plays a role, but I live in the US and in 2019 I already had tents in certain regions because of the flu because it was not a mild flu year at all. Aging is also less extreme here.
Here, however, the drug epidemic plays a huge role, because the same cartels that smuggled more than 20 million people into the country in 2021-2024 also transported tons of drugs. Now that the Abominable Orange Man has been re-elected, both numbers have plummeted. Drug deaths and alcohol abuse among certain groups are issues during corona that have undoubtedly caused excess mortality here, but not in the Netherlands.
And I also see a discussion taking place in Japan, where there was virtually no abrupt effect and excess mortality after 2023 has indeed turned into undermortality, but 2022 and partly 2023 are therefore mystery years. In the UK, however, according to the official authorities, it is a result of outdated/less good statistical calculation methods and with the new 'better' there is actually no excess mortality at all (*), and in Germany it was the heat waves according to the government (**). So we see that there is an excellent explanation everywhere and that the peaks in 2021-2024 have absolutely nothing to do with the vaccine. But we also secretly don't see a coincidental dip anywhere in 2021-2023, and the special things all moved in one direction.
*) https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/causesofdeath/articles/estimatingexcessdeathsintheukmethodologychanges/february2024/previous/v1
**)
https://www.destatis.de/EN/Themes/Society-Environment/Population/Deaths-Life-Expectancy/_node.html
Good comments!
For me, that is also one of the reasons why I think that the excess mortality is not too bad. I also have my doubts about some ages. But the majority fall reasonably within the bandwidths.
Of course, this is only the case if you do not take the 'pull forward' effect into account, as I indicate elsewhere.
For the elderly, I think you should take this into account, and after the 2020 and 2021 peak years, we would expect an extra low mortality in 2022-2024. However, we don't see them.
For groups under 65, it depends very much on what you want to identify as the cause. If you expect the cause to be a German heat wave or a direct effect of vaccines, you do not expect a 'pull forward' and only a relapse to the norm in that cohort. If you expect more cumulative heart damage effects and cancer due to vaccines or perhaps external changing lifestyle habits such as drugs/alcohol, you expect the mortality to continue and a return to the norm would be striking and therefore an indication that it must be something different.
I also don't know what the willingness to take a shot is in the Netherlands, but here in the US large groups are still eager to get to the pharmacy. So if the vaccines play a role, one would expect excess mortality to continue in countries with a high willingness to take shots, and especially in the cohorts that take those shots.
This is what I was referring to: https://exdeaths-japan.org/en/graph/weekly
This is from the Japanese government itself, and is honestly better to my taste because the standard mortality of the Japanese government is drawn in it. Of course we have to trust them, but as I understood there is a more open discussion there than in the US and EU.
In any case, according to their own data, we see an upward trend due to an aging population, but excess mortality in 2022-winter 2023 and then undermortality. Exactly what you would expect with excess mortality in the older cohorts. But since Covid was over in 2022, so the question is what happened there?
Are the data for men and women correct in those figures about the age classes?
If approximately the same number of men and women are born, how can mortality among men be much higher in ALL age groups? Is it not true that the mortality of women, especially in the age groups over 80 years, is considerably higher than that of men? A mistake in the color scheme?
Simple, because they simply became “extinct†sooner. This is only possible if mortality is higher in (almost) all classes of men. If that were not the case, they would live to be the same age as women.
Naar aanleiding van dit antwoord vermoed ik dat de blauwe en rode lijnen in de grafieken niet slaan, zoals ik aannam, op de frequentie van sterfte van mannen resp. vrouwen in die (leeftijds)klassen maar op de frequentie van sterfte van mannen onder al de mannen resp. van vrouwen onder al de vrouwen uit die klassen. D.w.z. het gaat niet om de breuk:
(sterfte mannen in G)/ (aantal vrouwen in G + aantal mannen in G) en idem voor vrouwen;
maar om de breuk:
(sterfte mannen in G)/(aantal mannen in G) en idem voor vrouwen;
G is hier de leeftijdsklasse (en alles natuurlijk geprojecteerd op 100K).
In dat geval is het immers verklaarbaar dat de lijn van vrouwen nog steeds onder die van de mannen kan liggen in de 80-plus klassen. Er sterven weliswaar meer vrouwen, maar de (relatief kleine) mannengroep kan toch nog steeds een hogere sterftefrequentie hebben in die klassen.
Het probleem is wèl dat daardoor eigenlijk het aantal groepen 2x zo groot wordt, want feitelijk zijn er 42 groepen waarvoor een berekening/ schatting wordt gemaakt, en niet 21. De betrouwbaarheid speelt vooral een rol in zowel de laagste leeftijdsklassen (waar weinig sterfte is) als de hoogste leeftijdsklassen (waar relatief weinig populatie is). Vooral het laatste punt is zwaarwegend omdat ongeveer 55% van de sterfte boven de 80 jaar is. We hebben b.v. in de 80-plus maar liefst 4 mannengroepen (in totaal dus 400K in de projectie) en maar ca. 250K/ 350K mannen per jaar die in de werkelijkheid in 2010-2024 overlijden !
Het is jammer dat in de plaatjes niet is gekozen voor de andere, eerstgenoemde interpretatie, want die zou niet alleen de betrouwbaarheid hebben verdubbeld, maar ook de duiding van de cijfers en de grafieken een stuk eenvoudiger en transparanter hebben gemaakt. In die interpretatie zou je immers de breuken direct kunnen optellen tot een (totale) sterfte frequentie in een leeftijdsgroep. En vervolgens zou je, met een wegingsfactor van het aandeel van de.leeftijdsgroep in de totale populatie, ook kunnen doorrekenen naar effecten op het totale ‘ruwe’ sterfte cijfer voor de hele bevolking. Nu is dat allemaal niet zo makkelijk en vergt het meer tussenslagen en rekenwerk om zover te komen.
Mijn oordeel over de uitsplitsingen naar de 5-jarige leeftijds- en geslachtsklassen is niet erg positief. Het heeft per saldo weinig toegevoegde waarde, daarmee kom je niet veel verder. Vooral de uitsplitsingen naar geslacht lijken me overbodig. Jammer van al die inspanningen: hoog cijfer voor vlijt, laag cijfer voor kennisvergroting !
1. Daar ga ik ook van uit (gestorven mannen/levende mannen en idem vrouwen), maar dat moet Anton bevestigen.
2. Van 94 jaar zijn er nog steeds 10.000 mensen; van 99 jaar nog ca. 1500. Daarvan gaat ruim 30% dood volgens de grafieken in dit artikel. Dat zijn nog steeds serieuze aantallen. Terwijl 2/100.000 bij kinderen slechts staat voor 3 overledenen. Dus die sterfte van ouderen is niet zo “wiebelig†en dus onnauwkeurig.
3. De gemaakte uitsplitsing maakt juist glashelder dat de sterfte/100.000, ook bij oudere leeftijdsgroepen, nog nauwelijks oversterfte laat zien (t.o.v. een norm van 2019). En dat is een keihard onomstotelijk gegeven. Want er is (behoudens jouw tegenbewijs) geen betere definitie van oversterfte dan “meer sterfte/100.000/per leeftijdscategorie t.o.v. een normâ€. En over die norm heb ik met Anton en Herman een debat. Hun norm is m.i. vrijwel zeker realistischer, maar minder overtuigend voor de boze buitenwereld. Daarom kun je m.i. beter strijden met een norm die niet ter discussie staat.
4. De uitsplitsing naar m/v laat juist in enkele categorieën zien dat er een flink verschil is. Als je dat niet uitsplitst valt dat hele patroon totaal niet op.
Dus Anton heeft dus zeer nuttig werk verricht, waarvoor nogmaals alle hulde!
Bedoel je daarmee dat, voor een vergelijk, de leeftijdscohorten anders gevoegd zouden moeten worden? Dus, als vrouwen 5 jaar ouder worden dan mannen, zouden de cohorten man 50-54 gevoegd moeten worden met vrouw 55-59?
Nee, je moet juist zo weinig mogelijk inhomogene zaken samenvoegen, omdat dat het beeld versluiert. En dat is met die opsplitsing per jaar en m/v uitstekend gedaan.