Thread by @fascinatorfun

That Peston article about more vaccinated people in the over 40s getting Covid than unvaccinated?

He did have reason for saying it from the PHE report but the problem was he failed to explain why the percentages were likely unwittingly misleading
theconversation.com/covid-19-why-i…

There have been several threads on it that I will attach to the end of this one, but I think it is easier for some to read a short article that fills in the gaps. This one by @VictimOfMaths is a good one.

I think there is another factor that should be taken into account.

How many in the unvaccinated group were previously (tested + ) infected and therefore likely to have some decent immunity.

Where should they be counted if what you are measuring is immune function. Should they be in a separate group from both unvaccinated and vaccinated?

Even then there are likely to be a proportion in the unvaccinated group that have some immunity because they were infected but were not tested.

Several reasons for this.

1/. They could not GET a test in the early months. How many hundreds of thousands fell into that group?

2/. They were asymptomatic or pauci symptomatic or had symptoms known to be Covid symptoms (WHO criteria) but were not symptoms on the U.K. criteria list for eligibility for a test.

This is even more of an issue with DELTA where runny nose, sore throat, headache, fatigue are all common symptoms (esp in children and the vaccinated) not on the eligible criteria list.

3/ People who are antagonist to getting a test (“don’t believe in PCR”)

4/. Those who are disincentivized to getting a test. They will lose money - maybe their job- if forced to isolate, and they really cannot afford it. Those who REALLY want to go to an event (football, a festival, a wedding, a funeral) and don’t want to be prevented “I feel fine”

Anyway. A portion of people in all those untested subsets are likely to i) have had Covid and ii) a good portion of those likely to have some level of immunity.

So it would be handy to know when looking at VACCINE effectiveness to what extent we are comparing it to unvaccinated NO previous infection and unvaccinated WITH a previous infection as they are likely to be in quite different immune functioning states.

There may well be a portion of the unvaccinated but infected who think it unnecessary to GET vaccinated “because I have got natural immunity” and are currently not convinced by the suggestion that vaccine as well will provide broader and maybe more durable immunity.

Anyway. If we don’t separate out the unvaccinated/ infected from the unvaccinated/ no infections, and - indeed- the previous infected + vaccinated from no previous infection + vaccination we will have less understanding on vaccine effectiveness.

It will tend to make vaccine effectiveness look lower than it really is and also mean we are not keeping long term track of the durability of ALL types of immunity.

And that takes us back to the article.

How many people are unvaccinated (let alone unvaccinated + infected)?

We do not know. It is surprisingly difficult to keep a bang up to the minute list of whole populations eg

- People move or emigrate and this has not yet registered/validated in systems
- Coding issues.

- Students might be registered at their home (family) address AND at their student address
- People who are seconded/sabbaticals and temporarily register with a new GP or are based in another country temporarily

Then there are all those people who have died due to Covid and other causes.

It takes time for systems to catch up.

The two main recording systems - ONS (using Census and other data) and NIMS (GP/NHS systems) produce different numbers for this for the above reasons.

For ADULTS:

ONS estimates there are 44.5 million adults in England, while NIMS has a higher figure of 49.5 million.

That is a 5 million difference.

We know who is vaccinated.

We should know who is tested.

And who has tested positive and negative

But that is a big difference when it comes to the unvaccinated and measuring vaccine effectiveness

If you like graphs this one shows the potential difference

Brown for the fully vaccinated (the bit we should definitely know from NIMS data)

The bit we don’t know re UNVACCINATED: Red for NIMS, Pink for ONS

Add to that the fact we do not have clear number on how many in the red/pink unvaccinated group have had prior infection and therefore likely some immunity and you can see that we are likely under rather than overestimating vaccine effectiveness.

So here are some additional resources for those who like information communicated in different way or add further insight.

The excellent @TimHarford ‘s #MoreOrLess

bbc.co.uk/sounds/play/m0…

Concentrating on the denominator issue of the ONS v NIMS issue from @Mike_aka_Logiqx

https://twitter.com/Mike_aka_Logiqx/status/1436982536039772161

A neat illustration from @john_actuary of the knock on effect when looking at vaccine efficacy (even without taking into account unvaccinated but previously infected, so likely some immunity)

https://twitter.com/john_actuary/status/1437014019706085385

And, finally, how the issue of immunity from prior infection MAY be skewing the Israeli data on two dose vaccine effectiveness making it look worse than it really is

https://twitter.com/jsm2334/status/1438241536555331584

Sorry it is a long thread, but I wanted to 1/ bring several resources together in one place and 2/ point out that there is another factor that maybe should be considered - and counted- when measuring vaccine effectiveness.

This adds to the points raised in @jburnmurdoch thread yesterday on durability but also on the issue that natural infection is likely to play re POST vaccination breakthrough

https://twitter.com/jburnmurdoch/status/1438100718112628740

AND

THAT “if a VE study doesn’t make sure its unvaccinated control group have all never been infected, its estimate of vaccine-induced protection will be biased downwards because of the level of protection that previously-infected people have”

In responses @JamesWard73 has flagged a thread he wrote a few days ago incorporating some of these issues

https://twitter.com/jamesward73/status/1436016896617467904

And I have just spotted a couple of tweets incorporating @jburnmurdoch thread on both durability and taking into account unvaccinated but with prior infection seems to have become detached from the thread.

https://twitter.com/fascinatorfun/status/1438406126178086916