UK Professor: "Data on mortality by vaxx status are junk"

Followers of this blog have seen how we enthusiastically dived into the UK data to compare All-Cause-Mortality between vaccinated and unvaccinated people. The data from UK were detailed enough to gain new insights, also useful for understanding the Dutch situation. Epidemiologist Jaime Bojas alerted us to errors in the articles. Fortunately, this was not so bad methodologically, but we still have to be very cautious, simply because it turns out that the data is impure. Unless England still starts counting noses.

Complaints have already circulated on social media about the unreliability of the ONS data. They report absolute numbers of deaths and age-stratified data. Statisticians would like to be able to calculate with age groups in order to arrive at an average mortality rate, for example, to be able to compare age groups or to make comparisons with countries with a different age structure. Mortality in a country with a very old population and mortality in a country with a young population thus obtain comparable mortality rates, corrected for the age effect. The countries are compared as if the age structure had been the same. Although this is a fair comparison – it also makes information less visible that may be important, especially in the case of an age-discriminatory disease. The criticism concerned in particular the results of those statistical conversions.

These statistical conversions (corrections, standardisations) seemed to circumvent by keeping the age groups strictly separate from each other (compartmentalisation). Then you have to make more graphs; You can't summarize or average anything – but I didn't want to because it's the difference per age group that is interesting. And if you work with an isolated age cohort, you don't have to convert or "standardize age" anything else.

Within a 10-year cohort, there is a risk that the youngest five-year cohort has a different profile than the older one. Because the vaccination coverage is almost the same between the elderly and the young within the 10-year cohorts, no significant difference can arise within the scope of the vax/unvax comparison.

For each age group, it is known how many people have died in recent years and, moreover, what their vaccination status was. This is accurately recorded per day per age group. So you only have to compare those deaths in terms of vaccination coverage with the relevant age group (and you know exactly per group whether the mortality rate is higher or lower in percentage.

It is therefore not wrong in the method or in the calculation but in the vaccination percentage as it appears in the heatmap at the bottom of the page of "Source 2" is communicated. Unfortunately, this official government statistic is not correct because the number of vaccinations is very closely tracked, but people do not have a good picture of the number of unvaccinated people. This is called wrong 'denominators' or 'denominators'.

A 'denominator' or 'denominator' is part of a fraction. However, the most important mistake is already made when determining the counters, before there is any dealing.

The number of unvaccinated people is calculated by subtracting the number of vaccinated people from the population. However, there are large differences of opinion about the size of the population in England.

The number of unvaccinated people is simply calculated by subtracting the number of vaccinated people from the population.
O = P – V (Number of Unvaccinated People = Population Size minus Number of Vaccinated People). However, there is a big difference of opinion about the size of the population in England. This has a huge impact on the number of unvaccinated people.

In this document of the National Health Service describes what causes the differences. See also the heading "In brief".

Short

NIMS (National Immunisation Management Service), which registers vaccination figures, is based on population figures that they update weekly. You would say that that must be good. Nevertheless, everyone agrees that these figures are a aboutestimate. This overestimation is attributed to a number of causes that seem quantifiable to me and for which it could also be reasonably corrected – but I'm not going down that rabbit hole... It must be more complex than it seems.

ONS (Office for National Statistics), which keeps track of mortality, on the other hand, works with England population mid-year estimate. The most recent version is from June 2021. These figures are overall very likely - and for some cohorts demonstrably - a underestimate.

Result:

NIMS has 60 million English people while ONS comes in at 53 million.

The problem of this becomes clear when calculating the number of unvaccinated people. That results in differences. The number of vaccinated people is carefully recorded. However, the number of unvaccinated people is calculated by subtracting the number of vaccinated people from the population. This results in dramatically different graphs for different population sizes:

The population is too large. After deducting the vaccinated, too many unvaccinated people remain. As a result, the mortality rate in that group decreases. After all, there are just as many deaths in a larger group.

The population is too small. After deducting the vaccinated, too few unvaccinated people remain. As a result, the mortality rate in that group is rising. After all, there are just as many deaths in a smaller group.

In the left graph, mortality among unvaccinated people is shown too favourably. The purple bars should be higher, but we don't know how much higher.

In the graph on the right, something very strange is happening. This is because in some groups there are more vaccinated people registered than there are people(!). ONS therefore works with the indication 100%*, where the asterisk indicates that the vaccination rate is above 100%. Of course, this is not possible and the result is that, after deducting the vaccinated, a negative number of unvaccinated people remains. Hence those hanging rods. That data therefore falls through the cracks – which does not immediately mean that the NIMS data are useful.

What now?

In any case, the difference that the graphs from previous posts about England show is too detrimental to the vaccines. Maybe there is something to normalize this a bit (it is being considered) – but it won't be very precise... Which, by the way, also applies to the ONS approach, where the vaccines come out too positive.

Prof. Norman Fenton also paid attention to this, see the tweets below from last week. Also read the additions by O.S. who point to more major inaccuracies in the ONS report. He makes a case for the use of the NIMS data:

How policymakers, population and health statisticians have done their work there in recent years is puzzling. Before you know it, you will leave the EU on the basis of wrong figures! If anyone knows why the Population Register does not work in England (incl. migration data)

The following text will be placed above the existing posts:

Fortunately, we still have the figures.

The only hard numbers are the rough counts. If you organize them you will also see something, just look. More about this in a next blog.