Wednesday, 13 July 2022

All you need to know to understand why the ONS data on mortality by vaccine status is systemically flawed

You don’t need to look very hard into the latest ONS data on all-cause mortality by vaccination status to confirm what we have been saying for a long time (see for example here and here): the ONS data is so obviously flawed (due to miscategorisation, missing vaccine deaths, and underestimates of the proportion of unvaccinated) to be worthless. 

This simple extract from Table 3 on the age standardized mortality rate (ASMR) *** for non-covid deaths over the full period (Jan 2021 – May 2022) is all you need to know to realise everything else in the report is problematic.

 


age standardized mortality rate for non-covid deaths (per 100K people)

Unvaccinated

1474.3

Ever vaccinated

892.9

(note that the ONS have, curiously, not provided the combined population ASMR for either non-covid deaths or all deaths for this latest period - we have written to the authors of the report requesting these numbers since they cannot be computed from the data provided).

Since this is the rate for non-covid deaths it should - in both categories - be similar to the historical age standardized mortality rate in pre-Covid times. Well, it just so happens the ONS has produced exactly those figures:

Year

age standardized mortality rate

2019

925.0

2018

965.4

2017

965.3

2016

966.9

2015

993.2

2014

953.0

2013

985.9

2012

987.4

2011

978.6

Notice how stable those figures are year on year (even in the main 'covid' year 2020 the ASM only went up to 1043). So what could possibly explain a 60% increase in non-covid deaths in 2022 among the unvaccinated and a 5% drop in non-covid deaths among the vaccinated?

Even the most fervent vaccine supporters know that it is not a miracle cure for non-covid disease. Yet, if you believe the above ONS data, it must be - since the unvaccinated are dying at a much higher rate of non-covid illnesses.  We stopped doing detailed analyses of the ONS updates on this data in January when, having exposed the blatant flaws, we called on a) the ONS to publicly withdraw their dataset and b) the retraction of any claims made by others that are based upon it.  

In case there has been an error either in our understanding of Table 3 or in the ONS presentation of the data we will be happy to set the record straight and update this article. But the bizarre data in the above table is consistent with the bias we found in the previous ONS reports. This bias (whereby mortality rates for the unvaccinated are overestimated and mortality rates for the vaccinated are underestimated) will inevitably occur when people dying (of any cause) shortly after vaccination are classified as unvaccinated and where the proportion of unvaccinated is underestimated.

The above table exposes that, far from fixing the many problems (that we discussed in detail with the authors of the ONS reports) these systemic flaws may have got even worse. 

There was previously a laughable 'explanation' provided for the increase in non-covid mortality in the unvaccinated  - namely the 'healthy vaccinee effect'. We already debunked that both theoretically and empirically. In any case, even if such a ludicrous hypothesis was correct, it would still mean that all the data - and conclusions from it - would have to be adjusted to take acount of the bias (which they certainly have not done).

Anybody attempting to draw any conclusions from the ONS data, let alone do any intensive data analysis, are now wasting their time. While it is possible there may be some interesting trends to discover from the data (such as possibly noticed here), these are more likely to simply reveal the extent to which the misclassification problems and other flaws have changed over time.

One of the most depressing things about the last two years has been how silent almost the entire academic community of statisticians and mathematicians has been in the face of so many obviously flawed statistics/studies that promote the ‘official narrative’ on Covid and the vaccine. A small number have contacted us privately to say they see through the nonsense but cannot speak out because they know how damaging it will be to their careers. In contrast, some prominent statisticians and mathematicians, who have refused to call out obviously flawed data supporting the ‘Covid narrative’, have been vocal in trying to downplay/ridicule more reliable data that supports an alternative narrative. Perhaps some of the prominent statisticians might at least finally publicly acknowledge a problem with the ONS data since it is so obvious?

*** we have also complained about the fact that the ONS started to use the age standardized mortality rate (unnecessary when you have the detailed breakdown by age) since it is essentially just a model that may not be suitable in this context.

9 comments:

  1. Some old ONS data from october 2021 indicates a higher all causes death rate for the vaccinated , see the pdf at https://filebin.net/h0p60jo1hjmtz428

    ReplyDelete
    Replies
    1. I'm no longer allowed to write filebin links on twitter , strange ?

      Delete
    2. Filebin no longer available 🤔☹️

      Delete
    3. https://filebin.net/pnyafbdibucj14pz
      expires 6 days from now !

      Delete
  2. Office for Notional Statistics perhaps?!

    ReplyDelete
  3. Possibly a dumb question, but...have you considered the time period starts at Jan 2021? So ‘unvaccinated’ figures include the vast majority of the population during the pandemic peak.

    ReplyDelete
    Replies
    1. ie https://www.ft.com/__origami/service/image/v2/images/raw/https://d6c748xw2pzm8.cloudfront.net/prod/9eb92250-599e-11eb-b10f-f9a852556ef0-standard.png?dpr=1&fit=scale-down&quality=highest&source=next&width=700

      Delete
  4. Never use processed data (e.g. age standardized), always use raw data.

    ReplyDelete
  5. Isn't just statistical data that's worthless. Different countries employ wildly different definitions of "COVID infection, rendering the whole concept meaningless. A lot of efforts gone into forcing the word into our active vocabularies, "COVID-19" doesn't refer to any medical or biological reality, which is just a fancy way of saying there's no such thing. (See here: https://michaelthau.substack.com/p/smoke-and-mirrors-the-linguistic?s=w)

    ReplyDelete