Mortality · VITD-MORT-01 · v2
Borderline: within 4 points of the Doubtful/Refuted boundary, so a re-grade could land either side.
The two biggest trials built specifically to test whether vitamin D pills lower the chance of dying from any cause found no benefit: in a large US trial [1] and a large Australian trial [15] the vitamin D and placebo groups died at essentially the same rate. Trials in sicker groups — critically ill patients [8] and heart failure patients [9] — also found no survival benefit. The only places where deaths dropped were a short-term COVID trial that compared a high dose against a standard dose rather than against a dummy pill [19], and a modelling study of UK Biobank data that found a benefit only in people who were short of vitamin D to begin with [20]. Because the strongest and most direct evidence shows no effect, and the signals of benefit are confined to special groups or non-placebo comparisons, the panel judged the broad, unqualified claim unsupported.
The two largest purpose-built RCTs are null for all-cause mortality: VITAL [1] reported a hazard ratio of 0.99 (95% CI 0.87 to 1.12) and D-Health [15] reported 1.04 (95% CI 0.93 to 1.18), with D-Health's point estimate above 1 and an exploratory signal of higher cancer mortality. RCTs in defined clinical populations agree: VIOLET in critically ill deficient patients [8] found 90-day mortality of 23.5% versus 20.6%, and EVITA in heart failure [9] found a hazard ratio of 1.09 (95% CI 0.69 to 1.71). The only mortality reductions are a high-versus-standard-dose COVID effect in COVIT-TRIAL [19] seen at 14 days (adjusted HR 0.39, 95% CI 0.16 to 0.99) that was not sustained at 28 days (HR 0.70, 95% CI 0.36 to 1.36), and an observational emulation [20] confined to insufficient (HR 0.85 and 0.81) or deficient (HR 0.79 and 0.75) strata. Cancer-survival trials [12] and [14] showed no overall-survival difference. The contested point: The division is over population: when the UK Biobank emulation [20] is weighted to the actual trial populations the effect is null (HR 0.97 and 1.02), but restricting to vitamin D-insufficient or deficient people yields substantial reductions. Whether that deficient-population signal, which comes from observational emulation rather than randomisation, reflects a real supplementation effect is the open point the panel left unresolved.
All 24 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Vitamin D, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
The two largest RCTs built to test this exact endpoint — VITAL [1] (HR 0.99) and D-Health [15] (HR 1.04, point estimate above 1 with an exploratory signal of increased cancer mortality) — are null, so the best evidence leans against the claim rather than for it.
Every mortality reduction in the corpus is conditional and indirect: COVIT-TRIAL [19] is a high-vs-standard-dose comparison with no true placebo and a 14-day effect that disappeared by 28 days; the deficient-stratum reductions in [20] come from an observational UK Biobank emulation (residual confounding), not a randomized supplementation effect.
Where it remains uncertain. The only reductions are a short-term, non-placebo high-dose COVID comparison that faded by 28 days [19] and an observational emulation [20] limited to vitamin D-insufficient or deficient strata, leaving a deficient-population effect unexcluded.
Aleatory
In the general-population trials deaths were a small fraction of participants (D-Health [15] recorded 1100 deaths among 21,315; VITAL [1] 978 deaths), so chance variation around a near-null effect is wide.
Epistemic
No randomised trial in the corpus enrols a vitamin D-deficient general population for all-cause mortality; the deficient-stratum benefit comes only from the observational emulation [20].
Inferential
The only all-cause reductions in a general cohort [20] rest on a UK Biobank target-trial emulation, which is subject to residual confounding and is not a randomised supplementation effect.
The load-bearing papers from a corpus of 24, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Largest general-population vitamin D RCT (N=25,871): no all-cause mortality reduction.
“In the analysis of death from any cause (978 deaths), the hazard ratio was 0.99 (95% CI, 0.87 to 1.12).”
Panel auditInternal validity: high · scope of inference: Generally vitamin D-replete US men aged 50+ and women 55+ free of cancer and CVD at baseline.Key limitations: unscreened mostly replete population · mortality a secondary endpoint
Critically ill deficient population: no mortality benefit despite correcting deficiency.
“The 90-day mortality was 23.5% in the vitamin D group (125 of 531 patients) and 20.6% in the placebo group (109 of 528 patients) (difference, 2.9 percentage points; 95% CI, -2.1 to 7.9; P = 0.26).”
Panel auditInternal validity: high · scope of inference: Critically ill, confirmed vitamin D-deficient ICU patients at high risk of death.Key limitations: acute critical-illness setting · single mega-dose
Heart failure RCT with all-cause mortality primary: null, point estimate above 1.
“Mortality was not different in patients receiving vitamin D (19.6%; n = 39) or placebo (17.9%; n = 36) with a hazard ratio (HR) of 1.09 [95% confidence interval (CI): 0.69-1.71; P = 0.726].”
Panel auditInternal validity: medium · scope of inference: Patients with advanced heart failure and 25(OH)D below 75 nmol/L.Key limitations: small sample · advanced-HF population · signal of harm on MCS implants
In-hospital mortality was numerically higher with vitamin D (7.6% vs 5.1%) and not statistically different.
Panel auditInternal validity: medium · scope of inference: Hospitalized moderately-to-severely ill COVID-19 patients given a single high dose.Key limitations: small sample · short in-hospital window · mortality secondary
4 of 24 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Two large well-powered RCTs designed for this endpoint are null and the only mortality benefits are confined to deficient subgroups or short-term non-placebo comparisons, so the general claim is refuted while a deficient-population effect remains open.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
RCT in vitamin D-deficient populations. Every null trial enrolled largely replete participants [1], [15]; the only deficient-stratum benefit is observational [20], so the subgroup the claim might hold in has never been randomised for all-cause mortality. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A randomised, placebo-controlled trial recruiting only 25(OH)D-deficient adults with all-cause mortality as the primary endpoint and daily physiological dosing.
Open question: Does daily vitamin D supplementation reduce all-cause mortality in people who are vitamin D-deficient at baseline? Why it is still open: The replete-population RCTs [1], [15] are null and the only deficient-stratum benefit is from an observational emulation [20], not a randomised trial.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 26 Oct 2026 · protocol v0.19