What will be the steady-state vaccine effectiveness against death?

Something has been bothering me about the way we are calculating vaccine effectiveness (VE) against death. The simplest way to explain to someone VE is to compare the fraction of total COVID Deaths of people who are vaccinated with the fraction of the total population that is vaccinated. If, for example, 80% of the population was vaccinated and 80% of COVID deaths were among the vaccinated, we would say the vaccine has 0 effectiveness against death. If the fraction of COVID deaths among the vaccinated was lower than 80%, then the vaccine has some effectiveness… the greater the difference between the % of people who are vaccinated & the % of COVID deaths among the vaccinated, the more effective we would judge the vaccine to be.

What has been bothering me about the “official” statistics on this is that it is ignoring the aspect of time. When calculations are being made, for example, for a week in the UK, the COVID deaths in that week among vaccinated and not vaccinated are calculated and compared against the % of population vaccinated and not vaccinated in that week. However, I think all but the most ardent critics of these vaccines would acknowledge the vaccines are effective for some period of time. Certainly we wouldn’t expect someone that got their 2nd dose two weeks ago to be at high risk for catching and dying of COVID. The evidence all points to the fact that these vaccines lose effectiveness over time.

I contend that it may be instructive then to compare the fraction of COVID deaths among the vaccinated in a given week, n, to the fraction of the population vaccinated at some number of weeks in the past…. n-m, where we assume m is the number of weeks the vaccine provides good protection. Hopefully this line of thinking becomes clearer with some charts I will insert below.

Data Used

• UK National Flu and COVID Surveillance Report : Figure 61 provided weekly vaccine uptake by age group

• UK Deaths by Vaccination Status : Weekly COVID deaths by vaccination status. This is the report where they lump everyone under 60 in one age group, so I will be focusing on the 60+.

Vaccine Uptake vs COVID Deaths Among Vaccinated

The darker blue curves represent the % of each age group that has received two doses at a given time. The light blue curve represents the % of COVID deaths in a given week among the people who have received two doses.

What should be of interest to all is whether what the steady-state will look like.(i.e. the long term behavior). Possible outcomes:

1. The light blue lines level off (perhaps we are seeing that at 80+) and remain below the dark blue, indicating there is at least some vaccine effectiveness against COVID deaths.

2. The light blue line continues to trend up and eventually meets the dark blue, indicating zero VE.

3. God forbid scenario- the light blue eventually rises above the dark blue, indicating negative VE.

Case 2 is of primary interest to me based on the notion of waning VE. To explore further, we can look at the horizontal distance between the two curves.

In these three charts, I show what fraction of COVID deaths were among the vaccinated in Week 38 & then trace back to see at what point the same fraction of the population was fully vaccinated. So, for the 80+, 86% of COVID deaths in Week 38 were vaccinated and it was in Week 16 that 86% of the 80+ population was fully vaccinated. For 70-79, we see almost the exact same numbers, with the time frame for equivalent population vaccinated percentage moving to Week 17. For the 60-69, 65% of COVID deaths in Week 38 were among the vaccinated, and 65% of the population was vaccinated by Week 19/20.

Taking the three groups, we see that % of COVID deaths among vaccinated in Week 38 is the same as the % of the population that was vaccinated between 18-22 weeks prior. Where does this lead us? Well, if we had a hypothesis that the vaccine worked very effectively for ~4 months (16 weeks) and we know that typically a COVID death occurs 2-4 weeks after an infection, then the above could be consistent with a long term VE of zero 😬.

On a cheerier note, the 80+ COCID deaths curve does look like it could be leveling off. If that is the case, we may instead be reaching a steady state, positive level of VE. Let’s hope that’s the case.

***UPDATE: I was a bit unsure at how clear my ideas are coming through on this one, so let me try to add one more bit of explanation. Suppose we had the idea that if we had a vaccine within the last 22 weeks (~5 months), we are safe from COVID death, but if we had the vaccine more than 22 months, we would not be protected. We can then divide the vaccinated people into those that are protected and those not. It is my contention that it may be appropriate to compare the % of COVID deaths among vaccinated at a given point (red circle below) to the % of vaccinated, no longer protected (blue star below).

***UPDATE 2, Electric Boogaloo (for those that didn’t grow up in the 80’s, that’s a joke about this ): A comment below made me realize that something might be unclear in this presentation. Namely, it is important to note:

1. The vaccination % curves of the population are cumulative. In other words, when we look at a given week, the point of the curve is the total % of the population who have been vaccinated by that week.

2. The % of COVID deaths among the vaccinated is not cumulative. Rather, it is the % of COVID deaths in that week among people fully vaccinated.

Why do I think this is the right way to look at things? Suppose you narrow in a particular week’s COVID deaths and you find that say 75% of those deaths were among fully vaccinated people. When determining whether the vax is effective or not, you would want to know the total % of vaccinated population (i.e. cumulative %).

The remainder of the post is arguing that, in addition, you do not want to look at the cumulative % of population vaccinated in that same week, but rather some number of weeks in the past. I thought of another analogy. Suppose that when you received the vaccine, you got on a space shuttle and left earth for 4 months. At a given point in time, when I look at COVID deaths among the vaccinated & compare it to the population of folks who are vaccinated, I don’t think it’s correct to include the folks on the space shuttle. This analogy would be consistent with the “vaccines work for 4 months” hypothesis.