Reviews/Polyphenols and Antioxidants/Antioxidant supplements and mortality
Sup · POLY-SUP-01 · v2
Borderline: within 4 points of the Established/Likely boundary, so a re-grade could land either side.
Across every antioxidant trial in this corpus, people taking supplements did not live longer than those on placebo. Selenium at a high daily dose actually raised deaths over long follow-up [2], a vitamin A and beta carotene trial in pregnant women showed no benefit [3], and an antioxidant cocktail's early edge in men faded once the trial stopped [6]. Antioxidants given around surgery and in intensive care were flat or slightly worse [8][10][14], and a multivitamin tracked for roughly 35 years changed nothing for total deaths [9]. The one clear benefit, fewer gastric-cancer deaths with a vitamin arm [12], is specific to that one cause of death and does not speak to death from all causes. The evidence pulls in one direction, which is why the panel treated the claim as well supported while noting the mix of very different patient groups.
Every antioxidant randomised trial that reaches the outcome shows all-cause or total mortality null or worse, never reduced: high-dose selenium raised all-cause mortality (HR 1.59, 1.02-2.46) [2], vitamin A and beta carotene did not lower pregnancy-related mortality (RR 1.15 and 1.21) [3], the SU.VI.MAX cocktail's early benefit vanished post-intervention (RR 0.98) [6], perioperative vitamin C and N-acetylcysteine were flat (HR 0.98, 0.54-1.80) [8], and ICU antioxidants were flat-to-adverse (OR 1.09, 0.86-1.40 [10]; adjusted OR 1.2, 0.8-1.8 [14]). A multivitamin over nearly 35 years left total mortality unchanged (HR 0.96, 0.89-1.03) [9]. The only reduction is cause-specific gastric-cancer mortality with vitamin supplementation (HR 0.48, 0.31-0.75) [12], which does not bear on all-cause mortality. The contested point: The division is whether a directionally unanimous but heterogeneous set of special-population trials (critically ill, pregnant, elderly, dysplasia and surgical patients) can carry a universal all-cause claim, since the general-population all-cause RCT present in the corpus [5][7] reports no results. The cause-specific gastric-cancer benefit in [12] is the one countervailing signal, and the corpus does not say whether it moved all-cause mortality at all.
All 14 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 Polyphenols and Antioxidants, 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 universal claim rests on a heterogeneous patchwork of populations (critically ill, pregnant women, elderly, dysplasia patients, surgical patients) rather than one adequately powered general-population all-cause-mortality RCT; the general-population D-Health trial [5][7] reports no results in this corpus.
Evidential independence is overstated: [10] and [14] are the same REDOXS trial, and [1] and [6] both derive from SU.VI.MAX, so the number of genuinely independent datasets is smaller than the paper count.
Where it remains uncertain. Vitamin supplementation cut gastric-cancer mortality (HR 0.48, 0.31-0.75) [12], a cause-specific benefit whose effect on all-cause mortality is not reported in this corpus.
Aleatory
[2] (N=491, states it 'was not initially designed to evaluate mortality') and [8] (N=600) produce wide intervals, so individual null results carry real sampling noise.
Epistemic
The large general-population all-cause-mortality RCT [5][7] contributes only methods and a statistical plan, no outcome data, so the best-matched population is unobserved here.
The load-bearing papers from a corpus of 14, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Antioxidant (selenium) did not reduce all-cause mortality and increased it at high dose.
“the hazard ratio (95% confidence interval) for all-cause mortality comparing 300 µg selenium/d to placebo was 1.62 (0.66, 3.96) after 5 years of treatment and 1.59 (1.02, 2.46) over the entire follow-up period”
Panel auditInternal validity: medium · scope of inference: Older adults in a moderately-low-selenium country; high-dose selenium supplementation.Key limitations: small sample · single centre · underpowered, not designed for mortality · high-dose selenium specific
Two antioxidants failed to reduce all-cause mortality in a very large cluster RCT.
“Use of weekly vitamin A or beta carotene in pregnant women in Bangladesh, compared with placebo, did not reduce all-cause maternal, fetal, or infant mortality.”
Panel auditInternal validity: high · scope of inference: Pregnant women and their infants in rural Bangladesh; pregnancy-related and infant mortality.Key limitations: pregnancy-related mortality, not general all-cause · beta-carotene/vitamin A specific · developing-country undernourished population
Perioperative antioxidants had no effect on 1-year all-cause mortality.
“Antioxidants had a HR of 0.98 (95% CI, 0.54-1.80), p = 0.96 for all-cause mortality vs. placebo.”
Panel auditInternal validity: medium · scope of inference: Patients with cardiovascular risk factors undergoing noncardiac surgery; 1-year mortality.Key limitations: surgical population · modest size · significant oxygen-antioxidant interaction · short perioperative exposure
Antioxidant cocktail showed no durable total-mortality benefit; earlier benefit disappeared.
“For total mortality (n = 90), the RR was 0.98 (95% CI, 0.75-1.26) for men receiving antioxidants compared to nonrecipients.”
Panel auditInternal validity: medium · scope of inference: French adults; combined low-dose antioxidants; during- and post-intervention total mortality.Key limitations: reduction is men-only subgroup · no CI for the during-intervention effect · effect did not persist post-intervention · women showed no effect
4 of 14 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.
A directionally unanimous, multi-agent, multi-antioxidant body of RCT evidence showing no all-cause mortality reduction strongly supports this negative claim, warranting an Established tier tempered by heterogeneity.
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.
General-population all-cause mortality result. The claim is universal but the best-matched trial in the corpus reports no outcome data, leaving the general adult population unobserved [5][7]. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: Report the completed primary all-cause mortality analysis of the pre-registered general-population antioxidant/vitamin trial [5][7].
Open question: Does antioxidant supplementation change all-cause mortality in a general, non-clinical adult population? Why it is still open: The one general-population all-cause RCT in the corpus [5][7] reports only methods and a statistical analysis plan, no results.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19