We were greeted by good news yesterday. A new UK population study from the University of Oxford, based on the ONS Infection Survey, shows that in fully vaccinated people asymptomatic infections were down 70% and symptomatic infections by 90%. The Telegraph has the story:
In the first large real-world study of the impact of vaccination on the general population, researchers found that the rollout is having a major impact on cutting both symptomatic and asymptomatic cases.
Sarah Walker, Professor of Medical Statistics and Epidemiology at Oxford and Chief Investigator on the Office for National Statistics COVID-19 Infection Survey, said that Britain had “moved from a pandemic to an endemic situation” where the virus is circulating at a low, largely controllable level in the community.
The new research, based on throat swabs from 373,402 people between December 1st last year and April 3rd, found three weeks after one dose of either the Pfizer or AstraZeneca jab, symptomatic infections fell by 74% and infections without symptoms by 57%.
By two doses, asymptomatic infections were down 70% and symptomatic by 90%.
But is it all as it seems? I wrote last week about vaccine studies that have glaring issues that everyone, including the authors, seem content to gloss over. Sadly, the same appears to be true of this study.
Here’s one of the key figures. Look at diagram A in the top left. The dots represent the infection rate in seven different groups of people defined by how long before or after vaccination they are and whether they’ve had Covid before.
It starts at the top with the group of people who are more than 21 days prior to being vaccinated and who haven’t had Covid before (and who may not have a vaccine booked or even be eligible yet for a vaccine). This group is the baseline so is given the value 1, and the number of infections in other groups are compared to this as a proportion. So the next group are those people who are less than 21 days before their first jab and who haven’t had Covid before, and they had 0.28 of the rate of infections that the first group had (once adjusted for various confounding factors such as location, age and sex).
This is the first oddity. Why do those less than three weeks before their first jab have around a quarter of the infections of those more than three weeks away from their jab? What is it about crossing that three-week threshold that has such a massive impact on infection risk, by far the biggest effect in the study?
The authors do offer a brief explanation, putting it down to “changes in behaviour due to either receiving the vaccination invitation letter or knowledge that individuals from their age or risk group are about to get vaccinated in their area”. But they offer no evidence of this mass change in behaviour triggered by the approach of the vaccination, and the vaccine invitation letter includes no advice to make any new effort to avoid people. In any case, it means the headline finding of the study should probably have been that being less than three weeks before your jab cuts infections by 72% – even more than being fully vaccinated!
Whatever the explanation (which I’ll return to below), this effectively gives us a new baseline for what we see next. Which is something that has become very familiar from Covid vaccine studies: the post-vaccine spike in infections. The infection rate rises to 0.38 in the first week after the first jab and then 0.45 in weeks two and three, a 61% jump above the pre-jab ‘baseline’. Yet this worrying phenomenon, which appears consistently in Covid vaccine studies, passes once more without mention. Why are researchers so uninterested in this?
We then see the infection rate drop until it hits 0.3 after the second dose, which would be encouraging were it not higher than the 0.28 pre-jab ‘baseline’.
Another point to note is that during the post-jab spike the proportion of symptomatic infections versus asymptomatic infections increases (look at the blue and yellow dots getting closer together 2-3 weeks post-jab in diagram C in figure 3, above). Since asymptomatic infection is associated with immunity (see the blue and yellow dots further apart for the fully-vaccinated and post-infection categories), this is corroborating evidence that the mechanism causing the spike may be a depression in immunity, possibly caused by the reduction in white blood cells post-jab observed in both the Pfizer and AstraZeneca trials.
Why is the infection rate in the more-than-21-days pre-jab group so much greater (nearly four times higher) than in the less-than-21-days pre-jab group? The authors propose (without evidence) mass behavioural change, but I’d suggest it’s more likely to do with when the tests were done. The period covered by the study is December 1st to April 3rd. In that period vaccinations increased by about the same amount each day and the halfway point was around February 14th. This means the less-than-21-days pre-jab group came on average from much later in the period than the more-than-21-days pre-jab group, as most of the jabs occurred after mid-February. This is significant since according to the ONS (below) infections declined very quickly in the first half of February so that any group weighted towards the latter half of the study period would have a much lower infection rate than one weighted to the earlier part.
Another possibility is that it was to do with age, as the vaccinated, and hence the less-than-21-days pre-jab group, were mostly over-60s, which was a low prevalence age bracket in this period. It may be both.
The authors control for a number of confounding factors, including age, location and high-exposure occupation (e.g. patient-facing health care worker), so in theory they should have eliminated many of these biases. However, presumably not all of them, as something must explain the 21-day threshold drop.
A further confusion is where the claim in the Telegraph article that “by two doses, asymptomatic infections were down 70%” comes from, as the post-second dose asymptomatic dot (the blue one) is 0.51 or 49% down, not 70%. Perhaps the 70% was taken from the drop in overall infections for the fully vaccinated. But if so this is, at best, sloppy reporting.
The 0.51 figure for asymptomatic infections for the fully vaccinated is relatively high, and is identical to the figure for the asymptomatic infection rate among those who have had Covid before. It most likely reflects immunity, as asymptomatic infection is typical of the immune system working (as is infection with low viral load, seen in diagram B in figure 3 to be more common in the fully vaccinated and previously infected). But those who believe in the myth of asymptomatic transmission will likely worry about this.
As before, I’m not trying to suggest that the vaccines don’t work. We know they are effective at increasing antibody prevalence and this must presumably have a significant impact on a person’s level of immunity. But this study is not a good example of how to show that they work. It has the oddity of the 21-day threshold drop in infections, and appears to show the vaccines being less effective than being in the pre-jab group. This is likely because it doesn’t differentiate the vaccine effect from the drop in infections that occurred anyway in January and February. It also has, once again, the worrying post-jab infection spike, that we are still waiting for anyone in a position of authority or influence to acknowledge, let alone investigate.
In the meantime, we’re left wondering. How much of the pattern of Covid surges coinciding with vaccine rollouts in various countries is occurring despite the vaccination programme, and how much because of it? Until the question is properly investigated – which will mean governments releasing data on the vaccination status of all who have died of all causes – the troubling questions will remain.