Thread by @jsm2334

I have seen a lot of people obsessing with Table 5 of the 8/21 UK Delta report:

402 Covid deaths among 47,008 fully vaccinated for case fatality rate 0.86%

253 Covid deaths among 151,054 unvaccinated for CFR=0.17%

Do vaccines increase CFR 5-fold?
1/4
assets.publishing.service.gov.uk/government/upl…

They also split data out by age (<50/>50).

For older group, CFR for vaccinated (1.81%) is 3.3-fold LOWER than CFR for unvaccinated (5.96%)

For younger group, CFR for vaccinated (0.05%) is 1.5-fold higher than CFR for unvaccinated (0.03%), but there are only 13 deaths.
2/4

Another case of Simpson's paradox, since a confounding factor (age) is STRONGLY associated with both outcome (death) and exposure (vaccination status) given risk of death in old >>> young and vaccination rate old >>> young.
3/4

Hopefully people soon realize that given the dramatic age disparities in vaccination and in risk of severe outcomes, any estimate of vaccine effectiveness must adjust for age by stratification or some other approach or it will give very confusing and misleading results.
4/4

Here is a brief blog post on this:
covid-datascience.com/post/simpson-s…

Just for fun, to show how EXTREME Simpson's effect can be, look what happens if we change the 13 deaths in vaccinated <50yr to 2 deaths.
In this case the CFR is HIGHER for unvaccinated for both <50yr (4.2x) and >50yr (3.3x), yet LOWER for unvaccinated if unstratified (5x).
CRAZY!