The headline is broadly defensible, but the qualifications matter. Effect sizes vary by population, the strongest claims rest on shorter trials, and credible voices push back on how it's typically framed.
they actually went and then did an informative censoring analysis to see if indeed informative censoring was exclusively responsible for the results in the banister paper and it turned out it wasn't in other words even when they repeated that methodology they still produced the finding you would expect
Every Sunday: the week’s new conflicts and verdict changes — and nothing else.
Native comments, Twitter mentions, and Reddit threads about this claim — surfaced together so the conversation isn't fragmented across platforms.
Bookmarking — the dossier-vs-overview split is the right call. Most of the time I want overview; sometimes I want receipts.
Would love a "what would change this verdict" RSS feed. Sign me up if it exists.
the follow-up study was which was done by keys at all in 2022 uh basically sought to improve on the methodology of the banister paper and it did something quite clever which is it repeated the analysis um using a different uh patient cohort so it's a Danish um patient population cohort