See this update about this work.
The latest ONS report on deaths by vaccination status has serious anomalies (as in their previous report which we analysed in this paper). The most obvious one is that their analysis claims age standardized NON-COVID mortality rate is TWICE as high in the unvaccinated as the vaccinated:
This is not credible. If it were really true then this would mean either
a) healthier people got vaccinated and/or very sick people did not get vaccinated. But then this would mean that all ONS estimates of vaccine effectiveness are massively exaggerated due to the confounding effect of prior health status; or
b) the COVID vaccines are full of magic fairy dust that cures non-Covid illness and increases your life expectancy
The report does indeed claim that a) is true (Page 5). But this contradicts both the NHS Guidelines (which are explicit in ensuring that the most at-risk and vulnerable get priority for vaccination) and the statement on Page 1 of the report (as well as anecdotal evidence that even terminally ill patients got the vaccine early):
And this on Page 8 of the report:
We know b) is false. So, if a) is also false then, as we claimed in our previous article, there is miscategorisation in the data; specificially, many of those dying shortly after vaccination are being classified as unvaccinated.
But even if a) is true the effect is the same: without the necessary adjustments to the data that we applied in our article, the data is worthless in terms of evaluating the effectiveness and safety of the vaccines.
A colleague has produced an initial summary of the various anomalies in the report, and an in-depth look at the 40-49 age group here.
As the data is simply not credible, we believe this entire new report should be disregarded.
It is also important to note that, if the ONS released all the raw mortality data, then we would not need to rely on the 'age standardized mortality rate' estimates at all. An explanation of the age standardized mortality rate and also how it can easily mislead people is here:
Our previous paper (which we are in the process of updating):