OTOH this is not an isolated data point. We have: the booster push (why?), the veteran study, the PHE data. Something doesn't add up. For whatever reasons, the observed epi VE, after all the confouding factors are accounted by reality, is fading for both infection and severe infection, though VE agains severe infection is fading at a slower rate. To the point of J&J VE against infection (the weakest of the bunch) being almost gone in 6 months. I don't have a time machine, but I'm really interested to see how the Jan 2022 VE numbers hold up across the board.
Given OAS concerns (please re-read the dengue study excerpt I already quoted), near zero covid risk for kids, rapidly fading VE against infection and the general lack of long term data about covid & immune system (no time machines), what exactly is the rush to vaccinate kids against the original variant?! And why do we pretend we can just tweak vaccines to any new variant, when, while technologically we can concoct any protein we like, the immune system doesn't turn on a dime?
> These data should be considered in the context of vaccination status of the population groups shown in the rest of this report. The vaccination status of cases, inpatients and deaths is not the most appropriate method to assess vaccine effectiveness and there is a high risk of misinterpretation. Vaccine effectiveness has been formally estimated from a number of different sources and is described earlier in this report.
In particular, the reduction of death rates among non vaccinated over time may be caused by natural selection, i.e. the weak died already while the remaining ones have unusually strong immune system (aka survivorship bias [2]).
There may be other unaccounted confounders, too, such as a change of behaviour over time where the degree of change is dependent on the age group and/or vaccination status. For example, an apparent reduction of effectiveness for younger people may be caused by them going back to work/study, thus increasing their probability of getting infected, while older folks staying in care homes as before.
Funnily enough, both Simpson's paradox and uncorrected confounders can result in observed correlations that are opposite to the real causality effect.
And to clarify, I'm not saying your VE numbers there are wrong, just warning readers about the dangers of statistics.
[1] https://plato.stanford.edu/entries/paradox-simpson/
[2] https://en.wikipedia.org/wiki/Survivorship_bias