Originally Posted by
Klimax
Quote:
Findings: Vaccine effectiveness of BNT162b2 against infection waned progressively from 92% (95% CI, 92-93, P<0·001) at day 15-30 to 47% (95% CI, 39-55, P<0·001) at day 121-180, and from day 211 and onwards no effectiveness could be detected (23%; 95% CI, -2-41, P=0·07).
Hm, I would ignore that last part about no protection. P-value is high. (Null hypothesis cannot be dismissed)
It's really standard stat. speak when p > .05. The last part does indicate waning after 211 days with a point Ve of 23% but a large CI. For that matter, detecting no protection is never possible. At best one can determine that a benefit or harm is statistically likely. Actually determining no effect can't be done. So what the paper is saying is that the p value is to high to exclude no efficacy in the 95% sense. But it does show some efficacly is likely but with a broad range.
Also, these studies are observational with all the associated confounding issues. Are vaccinated people more likely to get tested because they are more health aware or are they less likely because they ascribe any symptoms to something other than Covid-19 since they are vaccinated? That's a big potential confounder.