Facet of the supplements cluster (nucleus Vitamin and Mineral Supplements for Disease Prevention).
This is the fracture decision specifically, and it splits three ways that the general supplement-null
does not resolve: vitamin D alone (null), vitamin D + calcium (a real but small hip-fracture benefit
that lives in the high-risk/institutionalized stratum), and the net-effect question — whether even that
benefit survives the cardiovascular and kidney-stone harms of the calcium supplement. The decision is
stratum-dependent, and for the community-dwelling replete person the whole-strategy ledger runs
against the supplement (Bolland et al., 2011; Kahwati et al., 2018; inferred from Yao et al., 2019).
The efficacy structure — three distinct answers, one gold SR-MA (Yao 2019)
Yao 2019 (gold SR+MA, JAMA Netw Open) runs the three arms in parallel and they do not give one answer:
| Arm | Any fracture | Hip fracture | Read |
|---|---|---|---|
| Observational 25(OH)D (11 studies, 39,141) | RR 0.93 (0.89-0.96) per +10 ng/mL | RR 0.80 (0.75-0.86) | status marker, confounding-prone |
| Vitamin D alone RCTs (11 RCTs, 34,243) | RR 1.06 (0.98-1.14) | RR 1.14 (0.98-1.32) | null (dose-constrained) |
| Vitamin D + calcium RCTs (6 RCTs, 49,282) | RR 0.94 (0.89-0.99) | RR 0.84 (0.72-0.97) | 6% / 16% reduction |
- The observational gradient is not the intervention. «an increase of 10.0 ng/mL in blood 25(OH)D concentration was associated with 7% lower risk of any fracture (RR, 0.93; 95% CI, 0.89-0.96) and 20% lower risk of hip fracture» (Yao et al., 2019) — but the RCT arms are the causal test, and vitamin D alone does not reproduce it.
- Vitamin D alone is null — but the null is dose-constrained, not proof D is inert. The 11 RCTs «did not
find a reduced risk of any fracture (RR, 1.06; 95% CI, 0.98-1.14) or hip fracture (RR, 1.14; 95% CI, 0.98-
1.32), but these trials were constrained by infrequent intermittent dosing, low daily doses of vitamin D,
or an inadequate number of participants» (Yao et al., 2019).
Median 25(OH)D difference was only 8.4 ng/mL; «doses of 2000 IU per day are required to increase 25(OH)D
concentration by 20.0 ng/mL» (Yao et al., 2019). Seven ongoing
higher-dose RCTs (mean 2094 IU/day) will settle this — a
G (needs aggregation). - Vitamin D + calcium reduces hip fracture — and calcium is doing the work. «found a 6% reduced risk of any fracture (RR, 0.94; 95% CI, 0.89-0.99) and a 16% reduced risk of hip fracture (RR, 0.84; 95% CI, 0.72- 0.97)» (Yao et al., 2019). The D-alone-vs-D+Ca contrast (both from one source, so a clean within-source comparison) isolates calcium as the active co-ingredient at these vitamin-D doses.
Where the D+Ca benefit lives — the institutionalized/high-risk arm (the resolvent)
The 16% hip-fracture reduction is not uniform — it concentrates in the older institutionalized. «the combined supplementation of calcium and vitamin D was associated with more extreme changes in risk of any fracture in the RCTs of older participants (ie, aged 80 years) living in an institution than those younger than 80 years living in the community (P for heterogeneity = .02)» (Yao et al., 2019), and «The risk reductions achieved in the RCTs of calcium plus vitamin D were somewhat greater in RCTs among older participants living in institutions and in RCTs that achieved greater differences in blood» (Yao et al., 2019) — preceded by the standing qualifier that «the 95% CIs indicated some uncertainty for these estimates». This matches the recommendation history: combined D+Ca «has been recommended for prevention of fractures in older adults living in institutions and in those with low vitamin D status» (Yao et al., 2019).
This is the deficiency-repletion vs enhancement split instantiated on fractures: benefit in the high-risk/institutionalized (older, often vitamin-D-deficient, low dietary calcium — the repletion arm), little-to-nothing in the community-dwelling replete (the enhancement arm) -> Deficiency Repletion vs Enhancement.
Yao vs Kahwati is a DISTINCTION, not a tension — the population differs, not the finding
Yao (D+Ca reduces hip fracture) sits beside the held Kahwati 2018 (USPSTF SR: D+Ca does not reduce fracture in community-dwelling adults). The reflex is to file a tension — does D+Ca prevent hip fracture or not? It does not survive the parameter table.
| Parameter | Yao 2019 (D+Ca, hip) | Kahwati 2018 (D+Ca, hip) | Same quantity? |
|---|---|---|---|
| Outcome | hip fracture | hip fracture | YES |
| Intervention | D 400-800 IU + Ca 1000-1200 mg/day daily (Yao et al., 2019) | D 400-800 IU + Ca 1000-1600 mg/day daily (Kahwati et al., 2018) | YES (standard-dose daily D+Ca) |
| Pooled effect | RR 0.84 (0.72-0.97), 6 RCTs, n=49,282 (Yao et al., 2019) | HR 0.88 (0.72-1.08), 2 RCTs, n~39k (WHI+Porthouse) (Kahwati et al., 2018) | direction + point estimate agree; CIs overlap |
| Population | INCLUDES institutionalized elderly (Chapuy Decalyos, age ~85) + community (Yao et al., 2019) | community-dwelling ONLY; institutionalized / deficient / osteoporotic / prior-fracture EXCLUDED (Kahwati et al., 2018) | NO |
Not-joined check (ii) fires — different scope/population, consistent once matched. The point estimates (0.84 vs 0.88) barely differ and the intervals overlap heavily; the only material difference is statistical significance, and that is driven by which trials each pool contains. Yao includes the institutionalized trials (Chapuy) where the effect lives; Kahwati excludes exactly those populations by design, leaving a community-dwelling residual that is null. Yao itself reports the effect concentrates in the institutionalized (P-het .02) — so strip those and Yao’s residual approaches Kahwati’s null. No tension is filed; this is a distinction resolved by population (Kahwati et al., 2018; inferred from Yao et al., 2019).
What Yao adds to the fabric (type-F, NOT independent E). The high-risk/institutionalized fracture-benefit
leg was previously held only second-hand — Kahwati reporting the 2014 Cochrane review («The fracture
benefits overall appear to be largely attributable to benefits among the high-risk populations»
(Kahwati et al., 2018)). Yao now supplies
direct, within-review evidence for that leg: a pooled significant 0.84 with an in-review
residential-status interaction. This is a refinement, not an independent witness — WHI/Jackson 2006
dominates both Yao’s and Kahwati’s D+Ca pools (Porthouse 2005 is also shared), so they are not independent;
no [E-independent] is claimed (Kahwati et al., 2018; inferred from Yao et al., 2019). The two arms
are now both anchored — enhancement-null (Kahwati/VITAL, direct) and high-risk benefit (Yao, direct)
-> Deficiency Repletion vs Enhancement.
The net-effect ledger — the fracture benefit must be netted against CV + stone harm
A supplement taken for bone can carry costs elsewhere; judge the complete strategy on net effect. Yao flags this itself: «concerns have been raised about the safety of combining calcium and vitamin D for cardiovascular disease and higher risks of kidney stones associated with calcium supplements» (Yao et al., 2019). The two harm arms:
| Outcome | Effect (Ca±D supplement) | Source | Certainty |
|---|---|---|---|
| Hip fracture (benefit) | RR 0.84 (0.72-0.97); concentrated in institutionalized | Yao 2019 | low SoE, high RoB trials |
| Kidney stones (harm, D+Ca only) | pooled ARD +0.33% (0.06-0.60), RR 1.18 (1.04-1.35) | Kahwati 2018 | moderate SoE |
| Cardiovascular — MI (harm) | RR 1.24 (1.07-1.45) Ca±D (Bolland); RR 1.25 (1.07-1.45) (Myung) | Bolland 2011; Myung 2021 | direction corroborated; not independent |
| CV — CHD (harm) | RR 1.16 (1.05-1.28), I2=0%, n=9 | Myung 2021 | pre-specified primary; low-RoB-driven |
| CV — MI-or-stroke / CHD+stroke composite (harm) | RR 1.15 (1.03-1.27) (Bolland); RR 1.15 (1.06-1.25), I2=0%, n=14 (Myung) | Bolland 2011; Myung 2021 | corroborated; WHI-dominant |
| Cerebrovascular (stroke) | RR 1.13 (0.97-1.31), I2=0% — null | Myung 2021 | harm is CHD/MI-specific, not stroke |
| All-cause mortality | null (RR ~1.04) | Bolland 2011; Kahwati 2018 | low |
- The CV-harm arm (Bolland 2011,
high, CONTESTED). «Calcium or calcium and vitamin D supplements increased the risk of myocardial infarction (relative risk 1.24, P=0.004) and the composite of myocardial infarction/stroke (1.15, P=0.009)» (Bolland et al., 2011); «Calcium supplements with or without vitamin D modestly increase the risk of cardiovascular events, especially myocardial infarction» (Bolland et al., 2011). Held as a counterweight, not a settled harm: it is subgroup-derived (WHI women not on personal calcium) and WHI investigators + later reviews dispute it — so the direction is credible, the magnitude and even the sign contested. Label at point of use. - A second pooled MA firms the DIRECTION — Myung 2021 (
high, type-F, NOT an independent witness). Myung pooled 13 double-blind placebo-controlled RCTs (28,935 participants) with the CV endpoints as pre-specified primary outcomes (not Bolland’s MI subgroup) and reproduced the harm almost exactly: CVD composite (CHD+stroke) RR 1.15 (1.06-1.25), CHD RR 1.16 (1.05-1.28), MI RR 1.25 (1.07-1.45) — each with I2 = 0.0% — calcium «significantly increased the risks of CVD» (Myung et al., 2021), the harm «about 15% in healthy postmenopausal women» (Myung et al., 2021); absolute +8.6/1000 (CVD) and +8.8/1000 (CHD) (Myung et al., 2021). Three features refine Bolland rather than merely echo him: (i) the signal is concentrated in the low-risk-of-bias trials (RR 1.15, n=8; high-RoB trials null) — «the trials with low risk of bias in at least five items showed a significantly increased risk of CVD» (Myung et al., 2021) — the opposite of a bias/confounding artifact pattern; (ii) publication bias was formally tested and absent (Egger p 0.81/0.81); (iii) stroke is null (RR 1.13, 0.97-1.31), so the harm is CHD/MI-specific. - But Myung is NOT independent of Bolland — this is type-F firming, not
[E-independent]. Myung re-pools Bolland’s constituent trials (WHI/ref24, Reid/Auckland refs 15,22,26, RECORD/Avenell ref25, CAIFOS, Lappe, Baron), used unpublished trial data obtained from Bolland’s own meta-analysis for 5 of 13 trials, and cites Bolland — its «findings are consistent with those from the previous two meta-analyses» by Bolland (Myung et al., 2021). So the two MAs share their data spine; two-MA agreement buys no independence lift (Bolland et al., 2011; inferred from Myung et al., 2021). The contest is narrowed, not closed: the composite significance still leans on the WHI reanalysis (the same subgroup that made Bolland contested) — «When Bolland et al.’s study in 2011 was excluded, calcium supplementation was marginally associated with an increased risk of CVD, while all the other leave-one-out sensitivity meta-analyses showed a significantly robust increased risk» (Myung et al., 2021) (CHD became non-significant without WHI). And the constituent trials all had CV as a secondary endpoint, which Myung concedes as its chief limitation. A genetically-independent route does point the same way — a Mendelian-randomization study found «a genetic predisposition to higher serum calcium levels increased the risk of coronary artery disease and myocardial infarction» (Myung et al., 2021) (Larsson 2017, reported via Myung; not held — see gap (b)). - The exposure is the SUPPLEMENT bolus, not food calcium — do not overgeneralize. Bolland’s effect is «independent of dietary calcium intake» (Bolland et al., 2011), shows no dose-response, and its proposed mechanism is «the abrupt change in plasma calcium concentration after supplement ingestion … rather than it being related to the total calcium load ingested» (Bolland et al., 2011). Food delivers calcium gradually, so this harm does not transport to dietary/dairy calcium -> Is the Food Category Doing Any Work, Dairy and Bone Health.
- Bolland’s own ledger. «treating 1000 patients with calcium or calcium and vitamin D for five years would cause an additional six myocardial infarctions or strokes (number needed to harm of 178) and prevent only three fractures (number needed to treat of 302)» (Bolland et al., 2011) — hence a risk-benefit profile he calls unfavourable. The population caveat is load-bearing: this NNT-302 fracture benefit is computed in his largely community-dwelling, healthy postmenopausal cohort — the enhancement arm, where Yao and Kahwati also find little fracture benefit. So the ledger is cleanest exactly where the benefit is weakest.
Decision — by stratum
- Community-dwelling, replete, no osteoporosis/prior fracture (the enhancement arm). The whole-strategy ledger runs against the supplement: no fracture benefit (Kahwati direct null; Yao’s community residual) + CV-harm signal (Bolland + Myung; direction now corroborated by a second pooled MA, magnitude still discounted for WHI-dependence) + kidney-stone harm (Kahwati, moderate SoE). The CV harm bites hardest exactly here: Myung’s cohort was 92.8% women, mean age 66, «healthy postmenopausal women» (Myung et al., 2021) — i.e. the community-dwelling enhancement arm — so the harm is measured in the same stratum where the fracture benefit is weakest, strengthening the against-supplement verdict. Ca±D supplementation for fracture prevention is not indicated here; the fracture lever in this stratum is exercise, not the pill -> Exercise for Preventing Falls in Older Adults (Bolland et al., 2011; Kahwati et al., 2018; inferred from Yao et al., 2019).
- Institutionalized / older / vitamin-D-deficient / low dietary calcium (the repletion arm). Daily D+Ca (800 IU / 1000-1200 mg) has a real, if small, hip-fracture benefit (Yao 0.84, concentrated here). The harm transport is uncertain — Bolland’s CV signal was measured in community-dwelling replete women, not this stratum — so the net-effect is more favourable than in the enhancement arm, but the kidney-stone harm still applies and the benefit rests on low-SoE, high-RoB trials. A defensible lever for this specific high-risk stratum; not a general recommendation (Bolland et al., 2011; Kahwati et al., 2018; inferred from Yao et al., 2019).
- Vitamin D alone (any stratum, for fracture). Null at the standard doses tested; the higher-dose ongoing-trial question is open. Not a fracture lever on current evidence (Yao et al., 2019).
- Establishing deficiency is a prescriber act (this person’s labs), out of the wiki’s scope — the move stays test, then replete the confirmed shortfall, not supplement broadly to prevent fracture.
Confidence + gaps
confidence: medium(held, not raised). The efficacy structure is gold (Yao SR-MA), and the CV-harm arm is now better corroborated — a second pooled MA (Myung 2021,high) reaches the harm on pre-specified primary CVD/CHD outcomes with I2=0% and a low-RoB-driven signal, moving it from contested; subgroup-derived toward direction well-supported. It is not raised tohighbecause (i) the D+Ca fracture benefit still rests on low-SoE, high-RoB trials with wide CIs; (ii) Myung is not independent of Bolland (shared trials + Bolland’s unpublished data) so the agreement is type-F not a second witness, and the composite significance still leans on the WHI reanalysis (leave-one-out); (iii) the net verdict remains a synthesis across non-commensurable outcomes. The robust core is the shape (D-alone null / D+Ca-in-high-risk benefit / enhancement-arm net-harm), not any point estimate.- Gaps (G): (a) higher-dose (2000+ IU/day) vitamin-D-alone fracture RCTs — 7 ongoing,
G (needs aggregation); (b) a WHI-independent clean-population CV-harm test of Ca±D — the signal is now corroborated across two MAs but they share the WHI data spine, so an RCT designed with CVD as the primary endpoint, or the held Larsson-2017 Mendelian-randomization confirmation, would settle direction independently (the genetic route is currently reported only via Myung, not held -> acquisition candidate); (c) the net-effect ledger within the institutionalized/deficient stratum (where benefit lives but harm transport is untested).
Self-critique [run 2026-08-19, Myung ingest — before commit]
- No fake tension (the crux call). Myung reaches the same quantity, same direction, near-identical
magnitude as Bolland (MI 1.25 vs 1.24; CHD+stroke composite 1.15 vs 1.15) — this is corroboration, not
a clash, so it is filed as type-F firming, no
[[tension]], nocontradicts:edge. The parameter match was verified cell-by-cell (exposure, comparator, outcome, effect, pool) before writing. - Independence denied — no laundered-E. Myung re-pools Bolland’s trials, used Bolland’s unpublished
data for 5 of 13 trials, and cites Bolland;
[E-independent]is explicitly refused and the confidence is not raised on “two MAs agree”. The one genuinely-independent route (Larsson MR) is marked reported-via-Myung and routed to an acquisition gap, not claimed as held backing. - CV harm not overclaimed. The firming is bounded: leave-one-out WHI-dependence quoted, the
secondary-endpoint limitation named, “CONTESTED” narrowed-not-closed, magnitude discounted. Confidence
held at
medium. The RoB-driven point is stated as evidence against an artifact, not as proof of harm. - Number fidelity + dual test. All Myung figures trace to chunk 01 (cite.py-verified spans). Myung
earns its
sources:slot via distinct extracted claims (primary CVD/CHD/MI magnitudes, RoB stratification, leave-one-out, absolute risk, MR mechanism) — no pad.
Self-critique [run 2026-08-06, before commit]
- No fake tension. Yao-vs-Kahwati is filed as a distinction on the parameter table: same outcome,
overlapping point estimates (0.84 vs 0.88), difference is population/inclusion — not-joined (ii). No
tensionminted, nocontradicts:edge. - CV harm not overclaimed. Bolland is marked CONTESTED / subgroup-derived /
high-not-gold at every use; the direction is presented as credible, the magnitude/sign as disputed — not asserted as settled harm. - Supplemental-vs-dietary guard held. The CV mechanism (acute serum-calcium spike) and dietary-calcium independence are quoted, and the harm is explicitly barred from transporting to food/dairy calcium.
- No isolate/food halo. The page does not read supplements are useless (D+Ca works in the high-risk arm) nor supplements are fine (net-harm in the enhancement arm) — the verdict is stratum-split.
- sources: dual test. All three (Yao, Bolland, Kahwati) carry distinct extracted claims on this page (Yao: efficacy structure; Bolland: CV harm; Kahwati: the community-dwelling null pole + kidney-stone harm magnitude). No pad.