Vaccine 33% effective against omicron... Or is it?
Vaccine 33% effective against omicron... Or is it?
The media is falling over themselves to report a random press release that dropped today. In doing so, they're already mixing up the facts.
The Washington Post, for example, are claiming that the study featured 78,000 positive omicron cases.
Needless to say, it did not. This was a test negative design. It was abundantly clear just judging by the number alone (78000 positive?this early on?) what kind of study design we were looking at.
I don’t want to read too much into a study based on a press release (basically like judging a study based on an abstract), but I just wanted to reiterate that test negative study designs1 are usually incredibly misleading depending on the circumstances on the ground. Remember when “researchers” found that the vaccine was 5x better than natural immunity? Yeah, test negative designs are a good way to sell magic beans.
And we have to consider the trend with covid is for researchers to overestimate VE with test negative designs. Some of the myriad of problems include who is most likely to test (the hyper paranoid after a new variant comes out?), tests aren’t even taking the same measurement half the time (rapid vs PCR vs PCR with different Ct, etc), and the fact that testing is not random (who has the best access to tests in SA? People with the best access to vaccines maybe?). But we could have a whole post about this.
I’ll take the results with a grain of salt until the actual study is released. You should too. And don’t be surprised if omicron has even worse VE than that.
This type of design is looking at the proportion of positive tests in different populations. So if 1/100 unvaccinated tests positive and 4/600 vaccinated test positive, then VE would be 33.3% despite more positive tests among the vaccinated.
Unpopular Opinion: If the phylogenetic analysis of Omicron persists that it is a related but wild-type coronavirus (no furin cleavage site) that had an isolated existence, diverging from the main line of variants 16 months ago, and had a recent release that was noticed simply because of a lot of related testing going on, THEN it would stand to reason that mRNA and AVV jab sequences based on 2 year old Wuhan sequence might be something that could induce some immunity. I'm not saying it is a good idea, not saying that vaxx is a bright move, just that in the sequence-recognition sweepstakes, an older sequence is perhaps likely to be more susceptible.
A very important problem in a test negative design for things like Vaccine Efficacy study is that if study enforces that the first test positive (or negative) is counted only and that person is now dropped from the study for future events (say, the case is later vaccinated) which censors adverse outcomes, and blows up the denominator in those who haven't had it yet.
In real life, people are often positive or negative at different points in time and they vaccinate in an around that event which might be in a causal pathway. We cannot take any study seriously that cannot represent real life as time-dependent covariates rather than enforced censoring of multiple outcomes for the same person.
At least, in the Waning of Pfizer study, I detected this problem consistently where we would see negative efficacy for infection initially but positive efficacy for severe disease and death. Which makes no sense.
Well, it looked at the supplementary material and of course, there was a sensitivity analysis that did exactly what I wanted to see (let a case be a control and vice versa like real life). The positive efficacy in the first weeks against severe disease hospitalization and death flipped to -15.8% in line with infection susceptibility (and biological plausibility, how could getting an injection instantly save your life, but can't protect from infection?).
That was a 30% turn around. I expect this 29% reduction to be a similar scam. If you test positive while unvaccinated then decide to vaccinate and have an adverse event, the test negative studies tend to censor the event and just count the positive case when unvaccinated, bloating VE against severe disease.
The most important thing is knowing the disease susceptibility by age group. If the virus is predictably attacking younger people, then all gains are lost if the mean age of hospitalization drops from 85 to 28.
Of course, little mentioned fact is that they had just opened up the vaccination of 12-17 year olds on October 20, 12 days before Omicron. Never mentioned ever.
Source for Qatar Study showing 30% change in "VE"- Supplementary Appendix page25 Table S9
https://www.nejm.org/doi/suppl/10.1056/NEJMoa2114114/suppl_file/nejmoa2114114_appendix.pdf
--
Another problem with Negative Case control studies - If the publisher of the peer reviewed journal doesn't like your results, they have enough time to ask you to "get more data". I saw this happen twice where shocking changes happened to the original dataset. In one case, completely new people added just enough to reverse the vaccine danger of severe disease and death for Delta variant in Qatar in first 14 days.
I knew that we are in for a long haul. That unrevised draft is still present on medarxiv. I am 100% certain that we will not get out of this mess unless people like you or others push back since researchers either under pressure or due to bias have completely been corrupted and are willing to look away until we can't look away anymore.