Thread by @JamesWard73

I know, I’m meant to be doing that modelling thread (it will come later this evening!). But first I got distracted by today’s release of data from @PHE_uk – thanks to the amazing @kallmemeg and the team there. The main point of interest in today’s report has been …

https://twitter.com/kallmemeg/status/1435954475567566852

… the release of data on cases, hospitalisations and deaths split by vaccine status. In particular Table 4 of the report has caused some consternation as it suggests case rates are higher in vaccinated groups than unvaccinated, for age groups 40-79.

There are some real reasons why we might expect vaccine effectiveness to have declined – including the impact of delta, and waning of immunity over time. But there are also a number of potential confounders and distortions here, including:

a) The impact of prior infection (which may be higher in the unvaccinated group)
b) Correlation between vaccination and willingness to be tested
c) Behavioural impacts e.g. vaccination makes people less cautious, or higher-risk occupations more likely to get vaccinated ...

d) The treatment of “unlinked” cases (which may be more likely to be unvaccinated)
e) The calculation of how many people are unvaccinated, which depends on having an accurate estimate of how many people there are in each age group.

Just taking that last point: the PHE analysis uses the NIMS system to estimate the unvaccinated population. But we know that system can be inaccurate e.g. due to people not telling their GP when they emigrate / move to a different area. So instead let’s use the ONS 2020…

…estimates of population in each age group to estimate the number of unvaccinated (this is a bit rough & ready, because I can’t match the dates precisely, but it’s close enough to illustrate the point). We now get a very different graph for case rates by vaccine status:

As you can see, we no longer have case rates being higher in the vaccinated groups (except maybe in the 80+). I’m not saying the ONS numbers are definitively right either, they will also be distorted by other effects, but in this case I suspect this is a bit closer to the truth.

So long story short, please don’t get panicked by the case rates in Table 4, there’s many reasons why they don’t give the full picture. But this data is still a very powerful resource, and will get more powerful as we get more releases and can track trends over time.

The data also tells us something important about hospitalisations and deaths. We can calculate hospitalisation rates and death rates per case from this, which are unaffected by the denominators issue. Here's the graph for hospitalisations:

and for deaths. Even if you do get infected after a vaccine, your chances of going to hospital or dying are very much reduced if you’ve had the vaccine. (note the hospital data for under-18s will be distorted as the only u18s to have 2 doses are clinically vulnerable).

thanks to many who have contributed to this, knowingly or otherwise, including @alexselby1770 @nicfreeman1209 @ValentinThoss @BarclayBenedict and one I can't mention (you know who you are!)