A former senior figure at the Office for National Statistics (ONS) has criticised his old employer for breaching the Code of Practice for Statistics over its biased presentation of data relating to the vaccines.
James Wells, a statistician who served as head of the ONS UK trade team until 2019, has written to the UK Statistics Authority objecting to a new report, published November 1st, which leads with the statistic that “the age-adjusted risk of deaths involving coronavirus (COVID-19) was 32 times greater in unvaccinated people than in fully vaccinated individuals”.
However, as James points out, this statistic uses data from January 2nd to September 24th 2021, which includes the bulk of the winter deaths at a time when almost no one was vaccinated. This massively skews the implied vaccine effectiveness, as a fair comparison would only include periods when a significant proportion of the country was vaccinated.
An earlier version of the same report was released in September using data from January 2nd to July 2nd, which I criticised at the time. It made a similarly sensational claim about the high number of deaths in the unvaccinated compared to the vaccinated, garnering headlines that just 1% of deaths in the first six months of 2021 (640 out of over 50,000) were in fully vaccinated people. This was technically true, but completely misleading in terms of what it appeared to tell us about the effectiveness of the vaccines. Neither the ONS nor the journalists reporting on the data took care to make this clear.
In fact, according to PHE data, around 70% of deaths in August were in the double-vaccinated, not 1% – though that by itself is also misleading as an indication of vaccine effectiveness as it takes no account of how many were vaccinated. An estimate of unadjusted vaccine effectiveness (VE) against death from the raw data suggests VE was around 70% in the over-80s in August (dropping to 58% in the most recent report), and around 90% in 30-39 year-olds. This corresponds to more like a two to 10 times higher risk of death in the unvaccinated, not 32 times (though the estimates may need adjusting for biases in the raw data; on the other hand, studies from Israel and Sweden have found similar drops in vaccine effectiveness against death within six months).
The new ONS report stresses that “age-standardised mortality rates (ASMRs) can be affected by differing characteristics of the population in the vaccination status groups and the changing COVID-19 infection rate”. It therefore warns that “the ASMRs do not show causal links between vaccines and risk of mortality”. However, that note appears halfway through and is unlikely to prevent the misunderstanding of the top-line figure, which is presented as though (and most people will assume) it is telling us something about vaccine effectiveness. After all, why else present mortality risk split by vaccination status? The implication the ONS wants us to draw about the vaccines from its headline statistic is clear, whatever its notes halfway down the page might say.
James Wells is right to object to this misleading presentation of statistics, apparently to serve a Government narrative. Let us hope the criticisms brought by him and others will bring about positive change in how the ONS and other official bodies present Covid statistics.