In a UK Biobank analysis of 412,020 adults with 31,197 deaths, seasonally normalized 25-hydroxyvitamin D [25(OH)D] levels below 5 ng/mL were associated with a mortality hazard ratio of 2.03 (95% CI 1.90–2.18) versus levels ≥20 ng/mL. Even moderate deficiency (5–10 ng/mL) carried a 34% excess hazard. Causal weighting methods estimated 10-year mortality of 9.7% in the severely deficient group versus 4.8% in the replete group. Associations extended to diabetes, myocardial infarction, stroke, dementia, diabetic retinopathy, and end-stage renal disease — but skin cancers showed the reverse gradient, consistent with UV exposure confounding.
This preprint has not yet been peer-reviewed and results may change. That said, its methodological ambition is notable: the team applied an eight-harmonic seasonal normalization model to correct for calendar-timing misclassification, used one-year delayed entry to reduce reverse-causation bias, and employed directed acyclic graph-informed overlap weighting to approximate causal inference. These refinements go well beyond typical observational vitamin D studies. The finding replicates earlier Israeli and US cohort results, strengthening the dose-response signal. The critical unresolved question remains: do randomised trials of supplementation translate these observational gains into actual mortality reduction? Large trials like VITAL and ViDA have been largely neutral for mortality, possibly because they enrolled participants who were not severely deficient. This analysis argues that future trials should target adults with nadir-equivalent 25(OH)D below 10 ng/mL — a group that may represent the highest-yield intervention target.