1. Some people are looking at today's Public Health England report and concluding the Case Fatality Rate for Delta positive cases is 6.5 times higher for vaccinated compared to unvaccinated. But this is an instance of Simpson's paradox as shown by this table
2. Colleague .@MartinNeil9 pointed this out. In both age categories the rate among vaccinated is lower but when aggregated vaccinated is higher. It's because a much greater proportion in the older group are vaccinated compared younger group & most deaths occur in the former
3. It's worth noting, however, that in the <50 age group there's little difference in fatality rate between vaccinated & unvaccinated. Also, worth noting the concerns I've raised generally about all studies into risk/benefits of Covid-19 vaccines here: probabilityandlaw.blogspot.com/2021/06/why-al…
4. For a short video explaining Simpson's paradox, see:
5. And a more complex example is here:
6. Finally, for those who’ve interpreted this thread as an argument in favour of the vaccine: it’s not. I don’t think a convincing case of benefits over risks has been made, especially for the under 50s. But it’s important to call out stats errors on ‘both sides’ of the argument
7. Oops here is the correct table with the aggregated deaths:
8. Also important to note that the data that is completely missing is that relating to deaths (or serious adverse reactions) of vaccinated v unvaccinated people who were NOT classified as having COVID. So no information on vaccine risks
• • •
Missing some Tweet in this thread? You can try to
force a refresh
1. It's 16 days since we submitted a 250-word response to .@TheLancet pointing out the potentially serious limitation in the article they published (5 May) on Pfizer vaccine effectiveness. Response is still 'with editor'.
3. One caveat I should add is that I no longer believe ANY conclusions that are based on results of PCR testing are credible (and yes - this applies to conclusions in our own work where we relied on PCR test results).
1. There was massive media coverage over the recent study (published in The Lancet) claiming the Pfizer vaccine was "95% effective". But there are issues with the study and its analysis which mean the claim is exaggerated. See: probabilityandlaw.blogspot.com/2021/05/import…
2. The analysis fails to adjust the vaccine effectiveness calculation for different testing protocols for vaccinated and unvaccinated people. For example, asymptomatic unvaccinated people continued to be routinely tested while asymptomatic unvaccinated people were not.
3. The majority of the unvaccinated who PCR tested positive (and hence were classified as a 'case') were asymptomatic.
1. More on the rarity of asymptomatics with #Covid-19. In the latest rounds of Covid testing of Premier league footballers and staff, there were 0 positives among 2,787 tested:
2. There have consistently been very low numbers of positives among footballers and staff, even though:
a) these are among the only people who have not had to obey social distancing rules;
b) they travel all over the country and Europe; and
c) they cover the whole of the UK.
3. Can anybody shed light on whether confirmatory testing of positive results is done for the footballers? I understand their samples are tested by The Doctors Laboratory, a private clinic (which has incidentally been accused of health and safety breaches tinyurl.com/44p468e7)
View of today’s anti-lockdown march from Waterloo Bridge. This is 45 minutes after the front and I reckon at least an hour still to come.
Of course it was a lot more than anti-lockdown. The unifying theme was opposition to the Government using Covid to justify ever increasing infringements of civil liberties.
1. Note the crucial difference in wording between eye-catching graphic which says "1 in 480 ...with COVID-19" and the simple text which says "1 in 480 ...tested positive for COVID-19". The graphic is the one which is the lie.
2. The most important lesson from our work analysing the Cambridge data is that, with a false positive rate of around 0.35% for asymptomatics, most asymptomatics who test positive do NOT have COVID-19.
3. In the period we looked at there were 43 positives from the pooled PCR testing, of which 36 were found to be false positives after a confirmatory tests.
1. Remember the May 2020 headline story "Black people are 4 times more likely to die from Covid-19"? Our article on why these death risk statistics were misleading has been published today in the Royal Statistical Society's Significance Magazine significancemagazine.com/701
3. What is clear is that the original figures, which were so widely seized on by the media, were exaggerated - as we originally said. And even the current figures are also likely to be exaggerated by failure to account for demographic changes since the 2011 census.