What does "Unvaccinated" Mean
Tracking the Wrong Denominator Part 1
A lot of people have been talking about the Public Health England (now UK Health Security Agency - that’s not at all an ominous name) Vaccine Surveillance Reports. They come out regularly and appear to show both the some of the success and failure of the vaccination program.
Graphs like this one get sent all over socials by the vaccine skeptical:
While these charts are sent out by the vaccine proponents:
The top chart appears to tell the story of a failing vaccination program with everyone over 29 more likely to catch COVID if vaccinated than non-vaccinated. The lower charts show the very familiar refrain of “Continue to show strong protection against hospitalizations and serious disease” we hear from public health officials all day, every day. But what is the truth? If we honestly want to know the truth, we have to be willing to ACTUALLY follow the data, wherever it goes.
Major Caveat - I am NOT a Data Scientist or Public Health Statistician. I know just enough to hurt myself. Please point out where I’m going wrong and I’ll be happy to correct it. I’m literally, Just a Guy.
Each report is a 4 week sliding window, giving a current snapshot of COVID in the England with well broken out categories from “Unlinked” ie they don’t know the person’s vaccination status to “2 Doses” (which is still fully vaccinated in the UK). Many have noticed an interesting trend. If you subtract the Vaccinated case rate from the Unvaccinated case rate, you get an interesting result:
Leaving aside the mostly unvaccinated under 18s, we can see there is a steady downward slope to the lines. Negative numbers mean that the unvaccinated are catching COVID less than the vaccinated. The downward slope (which is very consistent in each age group) means that the unvaccinated case rate is either falling faster than the vaccinated case rate or rising more slowly. And moreover the regularity allows us to make predictions. For instance we can predict that the 18-29 year old will go negative in a little under 2 weeks.
Similar graphs can be made for Admissions and Mortality - both appear to show a hump in the past and a move to the downside while still being entirely on the positive side of the graph, meaning vaccination is showing benefit, but they are clearly declining.
All of this makes the vaccine skeptic feel justified. But what if this is misleading because of something virtually every government on earth is failing to do.
Count and Categorize the Previously Infected
The “Unvaccinated” category contains some number of people who are not vaccinated, but still have immunity. The question is how many people is that? Well it turns out in the UK, it looks like it is quite a lot.
The UKHSA also puts out the Weekly Flu and COVID-19 Report which contains the results of serum surveys taken regularly in the UK. Report 41 says this:
The Roche S test is for antibodies from both infection and vaccination. The Roche N shows a much lower overall prevalence of around 20% (higher in some ages lower in other for an average of 17%) and is only from natural infection. But that means in England, the actually non-immune population has fallen into the single digit percentages.
The supporting data package for the Weekly Flu and COVID-19 Report includes a massive table on population sizes and vaccination numbers. With this and the Roche S Seropositivity percentages, we can break the “Unvaccinated” category into "Non-Immune” (NI) and “Prior Infection” (PI) (to avoid confusion I am not using the term Natural Immunity which is also NI). It turns out that under 60 the average is 10% of the “Unvaccinated” population is Non-Immune. Over 60 it is around 15%.
This means the denominator being used to calculate the “Unvaccinated” case rates is between 6x and 10x too big. All of a sudden we see case rates much more in line with the hospitalization rates.
From a Public Health messaging perspective, this is GOLD. You can show that vaccination confers HUGE benefits vs the unvaccinated instead of showing Vaccinated Case Rates that are HIGHER than “Unvaccinated” rates. So why are we looking at the wrong denominator?