Finally a Paper Admits Negative VE vs Omicron

Danish Study Obviously contains many weasel words as well.

Recently a pre-print went to showing true negative VE. Not highly negative CIs that still cross 0, no, they are entirely negative after 3 months.

Since Delta is effectively gone, we can pretty much ignore those columns. Omicron is what we are interested in. The article is very short and easy to read so I highly recommend everyone take the time to read it. These charts are buried at the end. I think what the author's say is almost as interesting as what these charts say.

The topline (two paragraphs in, because I'm verbose), is that Pfizer shows a -76.5% VE in the period 91-150 days after two doses. Moderna shows -39.3% in the same time interval. You can see the lower bound of the confidence interval cross negative after just 31 days, with the Pfizer shots providing only 16% protection after 30 days.

This paper should be read very carefully. They did a post-mortem on cases, looking at 5747 cases between November 20, 2021 and December 12, 2021. This looks at roughly 22 days worth of cases in relatively early days of Omicron. Of particularly interesting note is how they calculated VE for booster doses (days 14-44). Here are their words:

It's only 60+, but 93% of their cases were under 60 years old, but they didn't compare to the Unvaccinated, I suspect they didn't have any to compare against (and they had to be 60+ to be boosted at that time). But that means that within 30 days of boosting, they had a VE against 2 doses, which has a negative VE! I actually don't know the correct way to do this math, my gut says that it's 54.6% of -76.5 or roughly -38% VE. I'm am fairly sure that's not right though because in theory it should be possible to have an overall positive VE with a good enough effect, and my method doesn't allow that. I'd be interested in learning the correct way to evaluate this.

The point is, the two comparisons are apples and oranges and the boosted VE cannot be compared directly to the unboosted VEs according to their wording. And we should talk about their wording, because there are some weird statements in here, like this one

This is extremely odd to see as the conclusion to the paper (the table at the top of this article directly follows this statement), and remember, they compared the boosted results to the 2 Dose results to get VE. They are saying that if you were over 60 and boost, you are less likely to get Covid than someone who is a) over 60 b) double vaccinated and c) much more likely to get sick than someone who is unvaccinated.

Moreover, they attempt to discount the awful 90-150 day results by blaming them on changes in behavior and some singular super-spreader event.

So once again it appears they are attempting to accept all of the “good news” and hand wave away the bad.

The problem here is that the trend was very clear leading into the final months, the VE was falling precipitously and the whole picture use coherent. Also remember that the time period of this study was just 22 days. It isn't like it spanned many months with a lot of cases coming in very late.

The design seems built specifically to reduce the impact of case spikes at the end of the investigation period. The data tell a clear and consistent story, and the conclusion is entirely mismatched to the data. Why?

Obviously we must consider cognitive dissonance, but I'm a little bit more hopeful here. I think they are attempting to bypass the gatekeepers, but doing so in such an easily spotted way as so invite others to see the reality in their data. I do wonder if maybe they laid it on a little too thick though, but it's hard to know what a true believer can see and what they cannot. It will be interesting to see if they get published or not and if their gambit works or not.

I'm the meantime, we have some rather compelling data to consider.