Mendelian randomization estimates roughly threefold greater cardiovascular risk reduction per mmol/L lower LDL than statin randomized trials
7 events · 3 assessments · 3 decisions
Reassessed (no status change)
Triggered by subclaim 777c4696 (MR ~50–55% CHD risk reduction per mmol/L) moving to verified against primary sources (Ference 2012 JACC 54.5%; 2018 EAS consensus OR 0.46). This is the numerator of the threefold comparison, so its verification reinforces the existing verified reading without changing direction. Both anchoring premises (MR ~one half; statin ~one fifth) are now verified, and the ~2.5–3× ratio sits inside what the primary sources assert verbatim ("3-fold", "up to three-fold"). Kept status verified, nudged confidence 0.90→0.92 to reflect that both load-bearing premises are now verified rather than one being weaker. Re-recorded the argument evaluation ('Genetic vs trial per-unit estimates', holds) to anchor it to the current premise standings. The third subclaim (per-unit benefit rising with exposure duration) remains unassessed but is a supporting mechanistic explanation, not required for the arithmetic, so it does not gate the verdict. Importance left at 0.35 (notable, descriptive comparison, not seriously contested). No dependent notification: no material change to propagate.
Reassessed: still Verified
verdict confidence 0.90 → 0.92 · credence 0.95
Reassessed (house-style cleanup, verdict unchanged)
Triggered by a curator_change. The Curator resolved the structural tension I had noted: the near-counterpart "MR and trial estimates agree" is a distinct reconciliation/consistency proposition, not the negation of this claim, and it legitimately takes this claim as a SUPPORTS subclaim. No merge or edge change affects this claim, so no structural edits were made. The Curator directed that the reader-facing reasoning drop the closing sentence narrating the flag to the Curator (per §12, reader-facing text must not describe the system's own bookkeeping). I re-recorded the assessment removing that sentence and added a clean reader-facing assessment (encyclopedia lead) distinct from the audit reasoning_trace. Verdict unchanged: VERIFIED, confidence 0.9, credence 0.95. The claim is descriptive of what two literatures report (Ference 2012 JACC and the 2017 EAS Consensus Statement both state the ~threefold figure verbatim), so it is robust. Re-recorded the sole argument evaluation ("Genetic vs trial per-unit estimates", holds) to confirm it still holds; its premises are unchanged. No dependent notification: substance of the verdict did not change, only the reader-facing framing.
Reassessed: still Verified
verdict confidence 0.90 · credence 0.95
Structured and assessed
First pass. Decomposed into one 'for' argument grouping three subclaims: a new MR per-unit estimate subclaim (777c4696, ~50-55% per mmol/L, importance 0.3), the existing verified statin per-unit claim (d81f89e3, ~one fifth), and the existing duration-of-exposure mechanism claim (d1491505). Matcher confirmed the statin and duration claims already existed; the MR estimate was novel. Assessed VERIFIED (conf 0.9, credence 0.9): Ference 2012 JACC states verbatim '3-fold greater reduction' (54.5% per mmol/L MR vs ~20% statin), and the 2017 EAS consensus says 'up to a three-fold greater proportional reduction.' 'Roughly threefold' is well within the range the primary sources assert. Set importance 0.35 (notable, largely settled description of what two literatures report; contestation 0.25). Escalated to Curator a structural concern: near-counterpart claim 2a38fdcd ('MR and trial estimates agree') lists this claim as a SUPPORTS subclaim when the two are logically opposed.
Assessed Verified
verdict confidence 0.90 · credence 0.90
This is an accurate description of a well-established finding in cardiovascular epidemiology. Mendelian randomization studies, which use lifelong genetically determined LDL levels as natural experiments, put the per-unit effect on coronary heart disease much higher than statin trials do: Ference and colleagues (2012, JACC), pooling nine LDL-lowering polymorphisms, found about a 54.5% lower risk of coronary heart disease per mmol/L lower LDL, against roughly a one-fifth (≈20-22%) reduction in major vascular events per mmol/L in the statin randomized-trial meta-analyses. That ratio is about two-and-a-half to threefold, which the original analysis reported as a "3-fold greater reduction" and the 2017 European Atherosclerosis Society consensus statement described as "up to a three-fold greater proportional reduction." The standard explanation, not in dispute among the parties who report the figure, is that benefit per unit of LDL lowering grows with the duration of exposure: a lifetime of modestly lower LDL compounds into a larger proportional effect than a statin started in mid-life and taken for a few years. The comparison is thus not a contradiction between the two literatures but a consequence of exposure time, and it is one of the observations underpinning the case for lowering LDL earlier and for longer. The figure is approximate ("roughly"), and the exact multiple depends on which MR estimate and which trial pooling are compared, but the threefold order of magnitude is what the primary sources assert.
Claim entered the graph