The Cleanest Failure in Nutrition Science: What VITAL Found About Vitamin D

Most arguments in nutrition science are messy. The studies are small, the effects are marginal, the confounders are impossible to untangle, and everyone leaves believing what they arrived with.

Vitamin D is not one of those arguments. It is the rare case where the observational evidence was overwhelming, someone ran the definitive trial, and the answer came back clean. That makes it the single best worked example of how nutritional epidemiology can point confidently in the wrong direction.

The observational case: Schöttker B, et al. Vitamin D and mortality: meta-analysis of individual participant data from a large consortium of cohort studies from Europe and the United States, BMJ, 2014.

The trial: Manson JE, et al. Vitamin D Supplements and Prevention of Cancer and Cardiovascular Disease (VITAL), New England Journal of Medicine, 2019.

The evidence that convinced everyone

The observational case for vitamin D was not thin. It was about as strong as observational evidence ever gets.

A 2014 individual-participant meta-analysis pooled 26,018 adults aged 50 to 79 across eight prospective cohorts in Europe and the United States. Comparing the bottom fifth of vitamin D levels against the top fifth produced a pooled risk ratio for all-cause mortality of 1.57 (95% CI 1.36–1.81).

A 57% higher death rate. And critically, the association held across age, sex, season and country — the consistency that usually signals a real effect rather than an artefact of one population or one measurement method.

If you were designing a supplement to sell in 2015, this is the dataset you would want.

What happened when it was tested

VITAL randomised 25,871 participants — men aged 50 and over, women 55 and over — to 2,000 IU of vitamin D3 daily or placebo, and followed them for a median of 5.3 years. It was large, long, properly blinded, and powered for hard endpoints rather than surrogate markers.

Every confidence interval crosses 1.00. Not a small benefit, not a trend that needed more follow-up. Nothing.

A 57% mortality difference in observation became a 1% difference in trial, and that 1% was indistinguishable from chance.

Why the two disagree

The most likely explanation is the least satisfying one: low vitamin D is substantially a marker of poor health rather than a cause of it.

People who are chronically ill, frail, obese, immobile or housebound get less sunlight and have lower circulating 25-hydroxyvitamin D. Vitamin D is fat-soluble, so higher adiposity sequesters it and lowers serum levels independently of intake. Systemic inflammation lowers it too. Each of those things independently raises mortality.

So the vitamin D level was reading the illness. Correcting the number on the blood test does not correct the thing that produced it — in the same way that painting over a damp patch does not fix the roof.

The caveat that actually matters

One limitation deserves more weight than it usually gets, and it is not a technicality.

VITAL was run in a largely vitamin D replete population. Mean baseline 25-hydroxyvitamin D was around 30 ng/mL — comfortably sufficient by most thresholds. Only a small minority of participants were genuinely deficient.

So VITAL answers one question decisively and a different question not at all:

• Does supplementing people who already have adequate vitamin D reduce cancer, cardiovascular events or death? No.
• Does treating genuine deficiency help? VITAL cannot tell you. It largely did not enrol those people.

That distinction is the whole ballgame, and it is what gets lost when the trial is summarised as "vitamin D doesn't work". Deficiency is a real clinical condition with real consequences for bone health, and correcting it is not the same intervention as topping up someone who is already fine.

What VITAL did find

For completeness, because selective reporting is the thing this article is about: VITAL's secondary and ancillary analyses were not uniformly null.

An ancillary analysis reported a reduced incidence of autoimmune disease over five years in the vitamin D arm. Analyses excluding the first two years suggested a possible reduction in cancer mortality, though not cancer incidence.

These are secondary findings from a trial that missed its primary endpoints, which is exactly the category of result this publication normally treats with caution — hypothesis-generating, not confirmatory. The same standard has to apply when the result is one you like.

What this means in practice

Testing is still worthwhile, because genuine deficiency exists, is common in some populations, and is worth correcting.

Supplementing an adequate level to chase a longevity benefit is not supported. The largest and best-designed trial we have looked for that benefit specifically and did not find it.

And the general lesson travels well beyond vitamin D. A 57% mortality association across 26,018 people in eight cohorts, consistent across every subgroup, still did not survive a randomised test. Whenever you see an observational association presented as a reason to take something, this is the comparison to hold in mind.

This is general information, not medical advice. Vitamin D deficiency is a clinical diagnosis — talk to your doctor about testing and treatment.

Sources

• Schöttker B, Jorde R, Peasey A, et al. Vitamin D and mortality: meta-analysis of individual participant data from a large consortium of cohort studies from Europe and the United States. BMJ. 2014;348:g3656. doi:10.1136/bmj.g3656
• Manson JE, Cook NR, Lee IM, et al. Vitamin D Supplements and Prevention of Cancer and Cardiovascular Disease. New England Journal of Medicine. 2019;380(1):33-44. doi:10.1056/NEJMoa1809944
• Hahn J, Cook NR, Alexander EK, et al. Vitamin D and marine omega 3 fatty acid supplementation and incident autoimmune disease (VITAL randomized controlled trial). BMJ. 2022;376:e066452. doi:10.1136/bmj-2021-066452