Vitamin D and the Evidence Gap: What the Major Trials Actually Found

There is a version of the vitamin D story that goes like this: millions of people are deficient, deficiency causes cancer and heart disease, and supplementing generously will fix both. That version is wrong — or at least, the largest randomised controlled trial ever run on this question found that it is mostly wrong. Understanding exactly where the evidence landed, why it is more complicated than either enthusiasts or sceptics suggest, and what it means for different kinds of people is worth doing carefully. The trial at the centre of this is VITAL — the VITamin D and OmegA-3 TriaL — run out of Brigham and Women's Hospital. It enrolled 25,871 adults, mean age 67, and followed them for a median of 5.3 years. Half received 2,000 IU per day of vitamin D3; half received placebo. The dose was chosen deliberately: it substantially exceeds the national guideline recommendations of 600 IU for adults under 70 and 800 IU for those 71 and older. The question was whether this meaningful boost would translate into meaningful protection against the two biggest disease categories in older adults: cancer and cardiovascular disease. The primary results were clear. The vitamin D arm saw 793 cancer diagnoses versus 824 in the placebo group — a difference so small it was statistically indistinguishable from chance. For major cardiovascular events — heart attack, stroke, cardiovascular death combined — the pattern was the same: no significant reduction. The American College of Cardiology's summary of the trial put it plainly: 2,000 IU per day was not effective for primary prevention of cardiovascular or cancer events in healthy middle-aged and older adults over that five-year window. Two other large trials reinforced this picture. The ViDA trial, run in New Zealand, and the D2d trial, which focused on diabetes prevention, together with VITAL produced a combined dataset of more than 30,000 participants. The Nature Reviews Endocrinology synthesis of all three is unambiguous: supplementing vitamin D-replete adults — meaning adults whose baseline serum 25-hydroxyvitamin D already exceeded 50 nanomoles per litre — does not prevent cancer, cardiovascular events, falls, or progression to type 2 diabetes. The phrase vitamin D-replete is doing significant work in that sentence, and it is worth holding onto. But first, the secondary findings from VITAL, because they are genuinely interesting and they drove most of the headlines that followed the trial. Despite the null primary endpoints, VITAL reported a statistically significant reduction in cancer-related death of roughly 17 percent in the vitamin D group. When the researchers excluded the first two years of follow-up — a reasonable move, since cancers present at baseline would dilute any true treatment effect — the cancer mortality reduction reached approximately 25 percent. A secondary analysis specifically examining advanced or metastatic cancer found a hazard ratio of 0.83, with a 95 percent confidence interval running from 0.69 to 0.99 and a p-value of 0.04. Vitamin D-supplemented participants were about 17 percent less likely to develop advanced-stage cancer, and the result was statistically significant, if only marginally so. The effect was strongest in normal-weight individuals, where a statistically significant interaction was observed. These are not trivial numbers. Reducing advanced cancer incidence by 17 percent, if real, matters enormously. Commentators at CancerNetwork pointed out that the supplement is cheap and low-toxicity, and argued the signal warranted serious further study. That is a fair argument. The tension between the primary analysis and these secondary results needs to be understood properly. A trial pre-specifies its primary endpoints for a reason: to protect against the inflation of false positives that comes from testing many outcomes. When the pre-specified endpoints are null and secondary analyses produce positive results, standard statistical practice says treat those secondary findings as hypothesis-generating, not hypothesis-confirming. The principal investigator, Dr. JoAnn Manson, was careful about this, noting that the results do not strongly support initiating high-dose vitamin D for prevention in healthy patients who are already meeting their requirements. That qualifier — already meeting requirements — is the key to the whole debate, and it points to what is arguably the deepest methodological limitation in VITAL and in trials like it. VITAL enrolled generally healthy adults with baseline vitamin D levels above the deficiency threshold. That was a deliberate choice: the trial was designed to test supplementation in the broad population, not specifically in people who were deficient. But this design makes it structurally unable to answer what may be the more important question: what happens when genuinely deficient individuals are supplemented? Post-hoc analyses of VITAL did hint at differential effects by baseline status. There were also subgroup signals suggesting stronger benefits in African American participants, a group at elevated deficiency risk due to the effect of higher skin melanin content on cutaneous vitamin D synthesis. The Nature Reviews Endocrinology review acknowledges that post-hoc analyses have suggested some extra-skeletal benefits for individuals with vitamin D deficiency specifically. But these are post-hoc analyses of a trial not designed to answer that question. They shift the prior probability — they do not constitute confirmation. There is also a structural problem with fixed-dose randomised trials in nutritional science. Giving everyone 2,000 IU per day regardless of starting point creates heterogeneity that the primary analysis cannot capture. Someone with a baseline 25-hydroxyvitamin D of 30 nanomoles per litre responds very differently to supplementation than someone at 70 nanomoles per litre. Genetic variation in the enzymes that convert vitamin D to its active form adds another layer. A fixed-dose, population-wide RCT averages across all of this, and the average can be null even when a subset of participants experiences genuine benefit. This is not a flaw unique to VITAL — it is a structural feature of how these trials are designed, and it means such trials may be unable to definitively resolve the question for at-risk subpopulations. The evidence on who is genuinely at risk of deficiency is less contested than the debate over extra-skeletal effects. People with very limited sun exposure — through indoor lifestyle, high-latitude residence, or consistent full-coverage clothing — are at risk, as are older adults, because the skin's capacity to synthesise vitamin D from sunlight declines substantially with age. This is part of why guidelines push the recommendation up to 800 IU at age 71. People with darker skin pigmentation have reduced cutaneous synthesis at equivalent UV exposure. Those with obesity have lower circulating levels even with comparable intake because vitamin D is fat-soluble and gets sequestered in adipose tissue. People with malabsorption syndromes — Crohn's disease, coeliac disease, conditions requiring bariatric surgery — are at risk because intestinal absorption is compromised. And people who are institutionalised or have very limited mobility rarely get meaningful UV exposure. For these groups, the question of supplementation is somewhat separate from the extra-skeletal efficacy debate. Correcting frank deficiency matters for bone health — one of the more robustly established areas of vitamin D science, though even the effect on fracture prevention in community-dwelling adults is more modest than commonly assumed. The real-world consequence of severe, prolonged deficiency is rickets in children and osteomalacia in adults — these are not subtle outcomes, and they are preventable. The bone health story also needs careful contextualisation. Randomised trials of vitamin D supplementation in older adults have shown mixed results on falls and fractures. Some meta-analyses find a benefit at higher doses; others find no significant effect in populations already replete at baseline. The pattern is consistent with what VITAL found in other domains: populations already above the threshold see little added benefit. On megadosing specifically: the cultural enthusiasm for high-dose vitamin D — 5,000 IU, 10,000 IU, or more, popular among people pursuing aggressive preventive health strategies — has essentially no RCT support for benefits beyond what lower doses achieve, and it carries genuine, dose-dependent risks. Vitamin D toxicity produces hypercalcaemia — dangerously elevated blood calcium — which can cause kidney damage, cardiac arrhythmia, and calcification of soft tissues. VITAL at 2,000 IU daily found no significant increases in hypercalcaemia, kidney stones, or gastrointestinal symptoms over its five-year follow-up. That is reassuring for that dose level, in that population, over that timeframe. It does not extend to doses three to five times higher, where case reports of toxicity are documented and the risk-benefit calculus shifts considerably. The 2,000 IU dose used in VITAL represents a reasonable test of the hypothesis and appears safe. National guidelines sit at 600 to 800 IU. The VITAL dose was two to three times that. If 2,000 IU in a generally healthy, replete population produces null primary endpoints on cancer incidence and cardiovascular disease, there is no scientific rationale to push further into megadose territory expecting greater protection. The logic that if some is good, more is better has not held up for vitamin D in this body of evidence. Routine population-wide testing of 25-hydroxyvitamin D levels is not supported by the evidence as a cost-effective intervention in low-risk individuals. For the high-risk groups identified above — the elderly, people with malabsorption, those with very limited sun exposure — testing is clinically sensible because it guides correction of a real physiological deficit. Testing in a healthy, young, sun-exposed adult with no relevant risk factors is unlikely to change clinical management meaningfully, because the outcomes people hope testing will prevent — cancer, cardiovascular disease — have not been shown to be preventable through supplementation in replete populations even at doses above guidelines. For people in the risk groups, supplementation without prior testing is also reasonable in many cases, because the doses required to correct deficiency and the doses known to be safe are not dramatically different. The Endocrine Society and national guidelines provide dosing ranges for at-risk groups, and the safety data from VITAL supports the idea that moderate supplementation in the 600 to 2,000 IU range is unlikely to cause harm. What remains genuinely open is worth naming. Whether the cancer mortality signal from VITAL — that 17 to 25 percent reduction — will replicate in a trial specifically designed to enrol deficient individuals is unknown. The minimum baseline serum level below which extra-skeletal supplementation benefits become clinically meaningful has not been established. The subgroup findings for African American participants and for lean individuals with advanced cancer are intriguing and underexplored. And whether targeted repletion of genuinely deficient people — bringing them from, say, 25 nanomoles per litre to 75 — requires different dosing strategies than what VITAL tested is an open question. The mega-dose vitamin D culture was built on observational data showing that low vitamin D levels correlate with worse outcomes across an impressive range of diseases. Those correlations are real. What the large RCTs showed — VITAL most definitively — is that correlation is not causation in this domain, and that supplementing healthy, already-replete adults does not reproduce the effects that observational studies predicted. The extra-skeletal claims — cancer prevention, heart disease prevention, diabetes prevention — have not been confirmed in rigorous trials for the general population. What has not been overturned is the rationale for correcting genuine deficiency in groups at real risk. That rationale rests on documented physiological need, established bone health consequences of deficiency, and a safety profile at moderate doses that is reassuring. For a healthy adult with no significant deficiency risk factors, high-dose vitamin D supplementation in the expectation of preventing cancer or cardiovascular disease is not supported by the best available trial evidence. For adults with established deficiency risk — older age, limited sun exposure, malabsorption, darker skin pigmentation in low-UV environments — correcting deficiency through moderate supplementation remains clinically reasonable, even as the scope of what that correction will achieve remains an open scientific question. The distinction between those two groups is precisely what the evidence supports, and it is a more useful conclusion than either the enthusiast or the sceptic version of this story.