CBS ziet nog 5 jaar oversterfte…!?

Like announced Herman Steigstra started calculating and we started thinking further. He just posted the result steig.nl. I shortened his title (the only thing I still had a question mark about) because whether CBS sees 'excess mortality' and how much depends entirely on the 'expected mortality' they choose: their own expectation, that of the RIVM or the 'standard mortality' that we also use on the mortality monitor. We haven't quite figured it out yet in that regard. It is now clear what the RIVM does. That said, below is the article.

CBS sees another 5 years of excess mortality, but RIVM does not see that

Sometimes it takes a while for the penny to drop, but this penny drops with a huge bang. On December 15, 2023, CBS published a forecast with the title:Prognosis of deaths; gender and age, 2023-2070. The interesting thing is that this is indeed a forecast, the first for the year 2023, which was almost over at that time. So that's actually a lame one, because it's accurate.

The forecasts for the following years are more interesting. We are looking until 2040, because the forecasts become more uncertain as the years go by. Even for 2024, the prognosis is uncertain, but these figures do allow us to get inside the head of Statistics Netherlands and read their view on excess mortality. But first we have to take a trip to the expected value, or the baseline, which we will discuss laternorm mortalitywill mention.

The baseline

Both Statistics Netherlands and we have made statistical calculations based on the mortality figures from 2010-2019 in relation to the composition of the population. We will have an update on this in November 2023extensive articlewritten. Our calculated expected annual mortality differed very little from the annual figures that Statistics Netherlands published at the time and therefore gives both sides some confidence about the correctness of the assumptions.

CBS and we calculate this baseline in almost the same way. We calculate a mortality risk over time for each age as a straight line, which decreases very slowly over the years for each age. This is an acceptable approach for a limited time, but at some point the decline in this line will have to stagnate. After all, it seems unlikely that we will ever become immortal.

Mortality rate

The mortality risk per age determines the total expected mortality per year. By multiplying these probabilities by the number of inhabitants per age (we do this separately for men and women) and summing them up, we calculate the total number of expected deaths in a year. Assuming that this probability decreases very slowly, but the population size increases slowly, the overall result is that annual mortality increases slowly. Until 2020, this was between 0.5% and 1.0% per year.

Life expectancy

A frequently used parameter to indicate the expected age to be reached is life expectancy. Life expectancy is defined as:the number of years a person is expected to live. If the future mortality rate would remain the same for every age, then life expectancy would also remain the same. But if the risk of mortality decreases, life expectancy increases at the same time. For someone born in 2020, this was 80 years for men and 83.3 years for women, we read inthis publication from CBSfrom 2020 in which we encounter this graph:

We see here that life expectancy slowly increases with age, but the difference between men (yellow) and women (purple) remains constant at 3.5 years. In oneother article from CBSFrom 2014, this difference is actually becoming smaller: from 7 years in 1990 to 2 years in 2030. But anyway, life expectancy is getting higher, we are getting older, so the chances of mortality are decreasing.

We can also calculate this forecast ourselves using our own calculation model:

We must note that this graph (just like CBS) is based on extrapolating the decreasing risk of mortality. The actual calculations are shown for women. For men, the uncertainty will become so great after 2030 (men would live longer than women), that we have aligned this with that of women, who are therefore also uncertain. But it gives an indication of what we might be thinking about, not an absolute truth.

And now the CBS forecast

After all, that's what we started with. Until 2024, CBS was responsible for the weekly reporting of mortality figures. These were shown in a graph, where mortality was compared to the baseline with bandwidth. This is an example of the reporting at the time:

The thin blue line is the baseline, the expected mortality per week, calculated by Statistics Netherlands. The thick line is the actual mortality.

From 2024, this reporting was taken over by the RIVM, which took the opportunity to present the figures against the background of their own forecast. It was therefore no longer the value previously expected by CBS that was being compared, but their own forecast. If the actual mortality did not deviate too much, the weekly value was “Not increased”.

Now it gets interesting, because now CBS came on15 december 2023so also with a prognosis. This is especially interesting when we look at the forecastsafter2023. We have brought everything together in this graph:

The blue line is the baseline we calculated (marked “Standard mortality”), which is therefore virtually identical to the annual figures from Statistics Netherlands (orange line) until 2023. Statistics Netherlands appears to have already taken into account the expected under-mortality in the years after corona in their calculations.

In 2020, as a result of corona, we see the actual mortality (black) increasing by around 13,000 compared to the expected mortality. The excess mortality is indicated in a light blue color and has continued unabated after the disappearance of corona.

Er worden allerlei verklaringen voor gegeven, maar de enige verklaring die nog steeds niet wordt geaccepteerd, is die waar vaccinatie een rol speelt. Uitgestelde zorg zou het zijn, “iets met corona” en zelfs vergrijzing werd genoemd door wie niet goed met cijfers kan omgaan.

Ook zijn de RIVM-prognoses als een groene lijn opgenomen. Volkomen onduidelijk wat we hiervan moeten begrijpen. Het lijkt meer een goedbedoelde lijn om aan te geven met welke sterfte we rekening moeten houden. Deze prognose heeft dus geen signaalfunctie meer maar wordt nog wel zo gepresenteerd (zie ooksterftemonitor.nl).

But CBS's explanation suddenly sheds light on the matter. The red line is the forecast that CBS is now issuing and we will start in the right half. The most important thing is that the red and blue lines will be very close to each other from 2029 onwards. This means that CBS assumes that the mortality rate will approximately follow our model from that moment on. Statistics Netherlands even expects a slightly lower mortality than we might expect based on the standard mortality rate. The direction of the line between 2020 and 2024 was already the correct one, because it is still being followed according to this forecast from CBS.

Alles wat valt in de categorie “vergrijzing” of “einde daling sterftekans” kunnen we wat CBS betreft nu dus meteen wegstrepen: de langzame daling van de sterftekans wordt weer voortgezet. Blijft als officiële verklaring over: uitgestelde zorg, waarvan al lang is vastgesteld dat dit een minimale invloed heeft gehad op sterfte. En als allerlaatste de categorie “onverklaarde oversterfte”, want vaccinatie mag het niet heten.

Prognose vanaf 2023

The next thing we will look at is the CBS forecast for mortality in 2023. Almost all figures were known, so the forecast and actual figures are virtually the same. But what follows is intriguing. CBS expects a slightly higher mortality for 2024 (it looks like mortality will be even 4,000 higher by the end of this year).

But what CBS also expects is that we will not get rid of excess mortality after that. This will slowly disappear, spread over 2024 to 2029, so there are still 5 years of excess mortality ahead. CBS expects it, but provides no further explanation. As a watchdog for our health, the RIVM also provides no explanation, even though they have virtually the same prognosis.

The question is therefore whether we can call this deviation of an average of 13,000/year from the expected value increased. RIVM explicitly reports “Not increased” every week, not only emphasizing the accuracy of their own prognosis but also that the mortality level is not a cause for concern. The institute does not provide any further insight into what is going on. This amounts to covering up the excess mortality that we have continued to see since 2020.

Norm mortality

De verwarring rond prognoses en verwachtingen wordt er niet minder op. Moeten we uitgaan van de oorspronkelijk berekende baseline van CBS, die ze nu niet meer publiceren? Of die van het RIVM misschien en moeten we accepteren dat wat we waarnemen ook meteen maar de norm is en daarmee automatisch ook “niet verhoogd”?

CBS is de weg kwijten kan kennelijk niet kiezen tussen prognose (rood in de grafiek) en te verwachten sterfte (oranje). Het RIVM kan niet overweg met baselines en signaalwaardes. Kunnen we dan misschien de procedure gebruiken, zoals gepubliceerd inResearchgate, met uiteindelijk vrijwel dezelfde waarden als die van CBS, maar wel voor iedereen te reproduceren?

Op deze site is eerder een pleidooi gehoudenvoor het invoeren van het begrip “Norm mortality", just as we already know it for many other variables. Just like the BMI (Body Mass Index), which we use to provide a guideline for healthy body weight. This does not move with the average weight gain. This must then be a reproducible calculation method based on historical mortality figures, where temporary shifts in the image have no influence.

It does appear that with this prognosis, CBS has done justice - in the somewhat longer term - to accepting the originally expected mortality on the basis of mortality probability. After all, Statistics Netherlands' forecasts are again based on the trend in the originally calculated expected mortality.

An “unknown cause” has caused more than 60,000 additional deaths to date. With the forecast for the end of 2023, CBS has emphasized that ournorm mortalityis the correct basis for the calculations.

Conclusions

In order to obtain a good assessment of excess mortality, there must be a reliable long-term prognosis. Statistics Netherlands has issued a forecast for the expected mortality for the coming decades and from 2029 this forecast will again virtually correspond to the values ​​calculated from our own calculation model, thenorm mortality. The difference between the observed mortality and theNorm mortalityis then an objective measure for theexcess mortality.

Sudden deviations of theNorm mortality, of het nu gaat om incidentele pieken of meerjarige trendbreuken, zullen verklaard moeten worden én bewezen onherstelbaar zijn, voordat de Normsterfte hierop wordt aangepast. Hiervoor zijn criteria nodig want geleidelijke afwijking van een voorspelling is onontkoombaar en die zal bijstelling vereisen. Gezien de belangen, de beleidsimplicaties en ideologieën die hiermee gemoeid kunnen zijn, zou een onafhankelijke commissie in het leven kunnen worden geroepen die de overheid c.q. het ministerie van VWS hierin adviseert.