Peter Attia· MD
if you are screening a population that is of relatively normal risk a positive predicted value of 20% is very very good
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.
if you are screening a population that is of relatively normal risk a positive predicted value of 20% is very very good
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.
now we're talking about something where we're approaching one or two uh positive tests will ultimately lead to a cancer diagnosis that's that's potentially actionable so it's um I think sometimes when people hear 40% they say gee that means there's still uh um a fair amount of um uh people who are going to get a a a positive test meaning a cancer signal detected and ultimately not but again for screening tests that's incredibly high yield