A statistical fallacy: unvaccinated excess mortality in Italy and at Nivel

Who Kaplan-Meier voor Dummies has read, understands what the 'belly' of a Kaplan-Meier graph says. Recently I started studying Italian again¹Previously included in they hve table toegespeeld waarin zo'n zelfde grafiek werd getoond. Ook hier was de conclusie, net als bij Nivel, dat de vaccins geen sterfte hadden veroorzaakt en dat de oversterfte bij de ongevaccineerden zat. We bekijken eerst de Italiaanse aanpak en vergelijken dan de bevindingen met het Nivel-rapport.

Kaplan-Meier 1: Italy

Italiaanse studie: COVID-19-vaccinatie niet geassocieerd met een toename van sterfte of geselecteerde PVR-SAE's-incidentie.

Are we worried about nothing? Are vaccines absolutely safe? According to experts, this is evident from everything: the situation in New Zealand, for example (which I here with ChatGPT) and a thorough, peer-reviewed, academic research from Italy would be decisive. The data comes from the Italian demographic databases and databases of hospital admissions for the period from January 2, 2021 to July 31, 2022.

The researchers of that peer-reviewed article in Vaccines | Like Nivel, came to the conclusion that mortality among vaccinated people was no greater than among unvaccinated people. ²did not show an increased risk of all-cause death, non-COVID death, or any PVR-SAEs, as compared to the unvaccinated A glance at the Kaplan-Meier graph makes it clear that there may be something to be bargained for.

They have made corrections to a number of comorbidities, including Covid-19, but write at the end of the strengths and weaknesses:

"...er kan sprake zijn van verstoring door het Healthy Vaccinee Effect, waarbij gevaccineerde mensen naar verwachting minder risicovol gedrag zullen vertonen."³there may be confounding due to the healthy vaccinee effect, wherein vaccinated people may be expected to adopt less risky behaviors It's really like this, see dangerous. That didn't seem like a good explanation to me and my well-read friend confirmed that. Wat ze beide niet meenemen is dat zeer frêle patiënten en patiënten met een korte levenshorizon (denk in maanden) bij 'ongevaccineerd' worden ingedeeld. En ondanks dat krijgen we gewoon deze grafiek voorgeschoteld:

It is striking that this graph is shown in a report that aims to show that vaccination does not lead to mortality or serious side effects. Apparently, this cannot be achieved without leaning on the difference with the less healthy unvaccinated. That difference is interpreted as positive for vaccinations.

Note that the vaccine only works after two weeks, which makes it extra vulnerable for a while. We don't see that in the graph: it works immediately.

Pas na een maand of 7 zien we een stabiele verhouding in rechte lijnen. En die lopen parallel aan elkaar: geen enkel verschil! Maar dat eerste half jaar, die "HVE-buik", die maakt het verschil, omdat alleen de totalen na anderhalf jaar worden vergeleken. Dan krijg je Nivel-achtige uitkomsten. De conclusie van de Italianen ligt dan ook voor de hand:

"Concluderend werd COVID-19-vaccinatie niet geassocieerd met een toename van sterfte of geselecteerde PVR-SAE's-incidentie. Verder onderzoek is nodig om de veiligheid van COVID-19-vaccins op de lange termijn te evalueren."

Nogal wiedes als je het "Unhealthy Non-Vaccinee Effect" meerekent. Dan komen de gevaccineerden er altijd beter uit.

This study exploits the Unhealthy Non-vaccinee effect to prove the safety of vaccines. Something that the literature (e.g. in the excess mortality report by Meester/Jacobs) explicitly warns against.

"Further research in the coming years will be required to evaluate the long-term safety of COVID-19 vaccines."

You can say that again...!

Kaplan-Meier 1I: The Netherlands

Nivel-rapport: Oversterfte tijdens de COVID-19-pandemie - Onderzoek onder mensen met en mensen zonder COVID-19-vaccinatie

In the much discussed Level-ratio there are two measuring points: the first at 3 months and the second at 12 months after vaccination. The development per day or per week does not appear in the study. Forget? Not important?

However, the first quarter is analysed separately. Could they be expecting more deadly side effects in the first three months? Then you shouldn't have to correct them away with increased expectations. You'd like to find something like that right.

In most age groups, the Nivel researchers expect a slightly higher mortality in the first 3 months than in the reference period. In the vaccinated only a little higher and why that is we do not read, it will have to do with corrections and weighings.

For the unvaccinated, the mortality expectation is therefore higher than the reference period. This looks like an attempt to correct for the Healthy Vaccine Effect: removing the expected excess mortality in people who did not receive the jab, for example because their date of death was approaching. After three months, that period will be over, they may have thought. But is that so? They could have seen that in their data. And even then, instead of months 1-3 and 1-12, they might have been better off reporting month 4-12. Otherwise, they will unjustly take the HVE with them in favor of the safety of the vaccinations.

If you expect an HVE, you cannot detect possible mortality due to vaccination by comparing with unvaccinated people. This is only possible if the study population is carefully composed, which is simply not the case with observational research.

For the passage over time, we look at one of the Kaplan-Meier graphs (71-81 years) by Bram Bakker, chapter 6 in the Master Jacobs report.

With the dotted lines I have indicated where the lines are straight. In the first few weeks, actually on the first day, the difference is really huge, like Herman Steigstra has calculated here.

Only after about 1 year (second vertical dotted line) does the situation remain stable. Only then does a continuous mortality risk begin to emerge that you might have something to say about. But this is also where the research period of the Nivel report ends...

Nivel has looked at two snapshots: at 3 months and at 12 months (indicated by the dotted lines). It is only after 12 months that there is a continuous trend. Up to 12 months, the differences per day become smaller and that cannot have been caused by the vaccinations. The differences are also not in line with Covid mortality trends.

I copied the blue and green dotted lines upwards. The green does not coincide with the red after shifting, which we did see in Italy. The unvaccinated continue to die more here. This will have to do with the age cohort, because this is the case with people in their fifties, as we saw in the previous article.

The blue line almost coincides with the orange. In the first period, we have seen that the selection was based on health. It is very well possible that the people who did not die immediately were not as healthy as the vaccinated people. Nevertheless, mortality is exactly the same. This would mean that the vaccination has a negative effect on this group: healthy vaccinated people have the same mortality rate as the unhealthier unvaccinated people.

Nog opvallender is dat de 'buik' van Nivel véél dieper is dan die in Italië. In Italy, they didn't seem to be bothered by waiting times, non-consenters and other data pollution. Dat zou daar dus 'zuivere' HVE kunnen zijn. Aan de hand van die grafiek kunnen we mogelijk iets meer zeggen over de samenstelling van die buiken in de Nederlandse data, die hebben geleid tot de waanzinnige resultaten van het Nivel-rapport. Als we daar de 'zuivere' HVE van afhalen, wat blijft er dan over qua datavervuiling, met name door het wachttijdprobleem? Ook dat heeft Herman further elaborated.

In summary, the only thing Nivel has measured is what the Italian study is also based on: the Healthy Vaccine Effect that they had wanted to correct and on top of that the administrative chaos. This is how they come to similar conclusions.

Is there anything in the report about the HVE? Sure:

• On average, the vaccinated group is healthier than the unvaccinated group, for example because seriously ill people, who are about to die, no longer get vaccinated.
• This implies that the reason for not vaccinating may be related to a higher risk of death. However, even after correction for certain characteristics, we saw that excess mortality among the unvaccinated/unregistered was much higher than among vaccinated people. There are still many features that we have not taken into account.

As Maurice pointed out: for Obesity and smoking for example, has not been corrected.

En dan toch maar ijzerenheinig volhouden...

What do we think?

The fact that there are scientists who continue to work with this kind of data and this methodology in government-funded studies to prove vaccine safety is worrisome. It is a statistical fallacy: an unmistakable HVE is being turned to vaccine safety and the scientific elite is watching.

People who no longer received the vaccine at the end of their lives and died unvaccinated are included to show a lower mortality risk for vaccinated people, not to mention other health issues among unvaccinated people. This not only produces disinformation, but it damages everything that has to do with science, institutes and governments (and with healthcare).

P.S.

Nivel's expectations, as they have been reported, are exasperatingly sloppy. They are rounded to 2 decimal places. For example, a 3-month expectation for unvaccinated people (18-35 years) of 0.01% mortality is calculated. On the face of it, this corresponds exactly to the reference period on which this expectation is based: also 0.01%. But that is the rounded percentage of 15 deaths among 223,552 young people. That's 0.0067%. By presenting this percentage as 0.01%, they set the standard 50% higher than what has been observed. I couldn't find the justification for that. Could it really be carelessness? When it comes to such small numbers, you report per 100,000, not per hundred, right? We just have to hope that those roundings didn't end up in the calculations...

Of course, these are only small numbers... 0.01% would mean 22 deaths, not 15 deaths of students and young parents in this study. A difference of 7. But calculated for the entire Dutch population, this is a difference of more than 100 young people. I can't imagine that medicine has gotten away with that in the past. Careless, after Honesty the second core concept in the Guidelines for scientific integrity.

Footnotes

• 1
  Previously included in they hve table
• 2
  did not show an increased risk of all-cause death, non-COVID death, or any PVR-SAEs, as compared to the unvaccinated
• 3
  there may be confounding due to the healthy vaccinee effect, wherein vaccinated people may be expected to adopt less risky behaviors