How can a vaccine have a negative effectiveness?
Are they implying it will make people more susceptible in the long term? That makes no sense at all.
Because I don't think that is what going negative means in this context. And reveals the problem with lay people like us trying to "do our own research" and dig into studies beyond the abstract. This is the explanation of the two Figures in question
"Figure 2. Adjusted vaccine effectiveness (any vaccine) against symptomatic Covid-19 infection among 842,974 vaccinated individuals matched to equally number of unvaccinated individuals through 9 months of follow-up. To model the association between vaccine effectiveness during follow-up, restricted cubic splines were used with 5 degrees of freedom.
Supplemental Figure 1. Adjusted vaccine effectiveness (any vaccine) against Covid-19 hospitalization or death among 842,974 vaccinated individuals matched to equally number of unvaccinated individuals through 9 months of follow-up. To model the association between vaccine effectiveness during follow-up, restricted cubic splines were used with 5 degrees of freedom."
Do you feel confident that you understand what that means?
I think I have a sort of idea of what is actually going on due to the explanation of matched pairing in the document, and looking at the rate at which the number of valid pairings declined during the course of the study. Because once 1 person in a pair suffers a COVID infection, then they are eliminated from future comparison. For all vaccines, it started with (N=1,684,958), dropping with each 60 day period, a huge drop to (N=635,402) at the 121 day mark, continuing to drop until the post 210 day period where (N=239,822). Looking at just the unvaccinated group, per period and the number of events.
1st - 4,719 (15-30 days)
2nd - 8,908 (31-60 days)
3rd - 7522 (61-120)
4th - 399 (121-180)
5th - 161 (181 - 210)
6th - 62 (210+)
Total: 21,771 events
Do you see what I see? The huge drop off of infections by unvaccinated people during the 121-180 day period, coinciding with the huge drop in overall matched pairs. Did the unvaccinated suddenly develop special skills to protect them? I don't think so. This is I think, where behavior comes in. Unvaccinated people got infected early or their normal behavior didn't lead to infection, the remaining uninfected kept those routines up, and if they survived 121 days without infection, getting to 210 days COVID-free doesn't sound all that impossible, does it? Like the zombie apocalypse, you have to be smart (or lucky) about what you do, to survive the initial wave. The unvaccinated that made it past 210 days are really good at avoiding COVID.
Meanwhile in the vaccinated group, their total overall was only 6147 events. But at that 121-180 window vaccinated people had 820 events. More than the 399 in the unvaccinated group. So doing the standard calculation, just like your checkbook when expenses > income, it went negative. The remaining periods vax also outnumbered the unvax group 718 to 161 and 522 to 62, and so it looks even more negative as shown in the chart. Their individual behavior probably didn't change between the first 121 days and the last 100 either, so the change is likely due to the vaccine effectiveness. What I think the negative shows is that, in general, vaccinated people are interacting with more people, and therefore more exposure risk, than the unvaccinated who have managed to get this far without a COVID event. Vaxxed people were less concerned with COVID infection and thus got infected by COVID. But I'd still rather be part of the 6147 event cohort than the 21,771 event one.
Consensus opinion has been growing around the initial dosage, especially for Pfizer, being inadequate against Delta. This is just evidence of that. I don't think this is a poor study, it's just we don't really understand what we are looking at. I also don't want study authors to have to change their methods because people are looking at the pictures and studies and not understanding what they mean, when their colleagues and intended audience understands just fine. We are lucky to be able to access this stuff as is. If we keep abusing and misinterpreting results they'll probably stick things behind paywalls and login screens. I want people to recognize what they don't know, and wait for more qualified people to make statements about what it means instead of jumping to none of it works!
