Primary Clinical Outcomes For A Single Emergency Department 2017 - 2021 (part 5)

SMLR modelling of 1.9 million admissions records for the emergency departments of an undisclosed NHS Trust: the healthy vaccinee effect revealed

Back in part 4 of this series I ended with the question… are vaccinated folk different? They’re certainly different once they concede to experimental gene therapy but were they different before they got jabbed?

Sometimes known as the ‘healthy vaccinee effect’ or ‘healthy vaccine bias’, I’ve tackled this subject before when I analysed something I called PROD (prior risk of death) and discovered that unvaccinated folk were sicker than vaccinated folk prior to death. Today I’m going to look at this key issue again but in terms of predicted risk of treatment within the ED and predicted risk of post-admission hospitalisation.

The good news here is that the data dump provided dates of vaccination for dose 1, dose 2 and the first booster, which means it was possible to flag all possible scenarios for the sample of 1,530,522 adult admissions over the period January 2017 – September 2021. Here are those scenarios and their frequency of occurrence:

I’m hoping these categories are self explanatory. What we have here is a time machine whereby we know what each individual eventually decided to do when it came to getting the jab. An obvious first comparison to make is between the unvaccinated (control) group and those who were admitted who would then go on to get the first dose some time later. This rules out the effect of the vaccine and enables us to compare their basic health status in terms of risk scores for the two primary clinical outcomes. If we tighten selection to look at adult admissions prior to vaccine rollout then the table reduces to this:

These are people being admitted to the ED over the same time frame (2017/w1 – 2020/w49). At this point in time nobody had got jabbed but 839,554 of 1,283,184 (65.4%) were going to get jabbed shortly after. Let us then look at summary stats for the two primary outcome risk scores:

‘Elo, ‘elo, ‘elo, wot’s all this, then? I’m going to bet that your eyeballs are seeing what my eyeballs are seeing and that is the mean and median scores for both risk of treatment and risk of hospitalisation are lower for the unvaccinated sample who would eventually go and get jabbed. That’s the healthy vaccinee effect right there!

In terms of factor ratios based around the median score we’ve got unvaccinated folk who remained unvaccinated after rollout yielding a factor of x1.03 for risk of treatment and x1.19 for risk of hospitalisation when compared to folk who would go on to get a jab when it became available. OK, so that 3% hike for treatment might not pass muster for some readers but a 19% hike for hospitalisation is hard to ignore.

What I think we need at this point is a pair of colourful slides to hammer the point home. Have a taste of these:

There’s no getting away from the fact that those who eventually decided to get jabbed during rollout were healthier than their counterparts who decided not to get the experimental therapy, both in terms of their predicted risk of treatment within the ED and in terms of risk of hospitalisation following admission. This is one of those awkward facts that the pro-vaccine brigade try to smother.

We should also note how complex the situation is, with no healthy vaccinee effect showing for the youngest age groups and a diminished effect for the oldest age group. If you want to tease this effect out to the max you’ll need to devote your analysis to those aged 60 – 89; beyond this range and the effect becomes diluted. Please bear this in mind when digesting various studies.

Tea & Crumpets

So there we go, a billion dollar question answered for the price of a pot of tea and four buttered crumpets. There is no doubt in my mind that vaccinated folk were healthier, and possibly more health conscious in the first instance; and unscrupulous investigators no doubt are very much aware of this when they press ahead with studies apparently showing vaccine benefit. What these despicable studies are going to reveal is nothing more than a confounding factor we might call basic health status, this being far more complex than a bunch of common-or-garden demographics such as age and sex.

With this knowledge on the kitchen table it is entirely possible to have a stab at assessing the impact of vaccination on risk of treatment and hospitalisation during the post-vaccine era, but only if we use these unbiased risk estimates as a propensity score in the first stage of a staged multivariate logistic regression.

I shall now go think some more, tug my beard and cogitate…

Kettle On!