Vitamin D and Mortality: What Trials Show

People with low blood levels of vitamin D die earlier than people with high levels. That observation held up across two decades of population studies, and it made the obvious next step look easy: give people the vitamin, watch the risk fall. When the VITAL investigators launched their trial, they noted that randomized data on vitamin D and cancer or cardiovascular disease were still limited. Those data now exist, and they did not go the way the cohort studies implied.

What an association could not settle

Low vitamin D travels with a long list of other things: frailty, obesity, chronic illness, and not going outside. Each of those independently predicts dying sooner. A blood level that drops because someone is already sick will look like a cause of illness in any dataset that only measures the correlation.

That is the reason the trials were built. An association tells you the two things move together. Only randomization tells you whether moving one moves the other.

What the randomized trials found

VITAL randomized 25,871 US adults (men 50 and older, women 55 and older) to 2000 IU of vitamin D3 daily or placebo, in a two-by-two factorial design with marine omega-3s, and followed them for a median of 5.3 years. Its two primary endpoints were invasive cancer of any type and major cardiovascular events.

Endpoint	Result	95% CI
Invasive cancer (primary)	HR 0.96	0.88 to 1.06
Major cardiovascular events (primary)	HR 0.97	0.85 to 1.12
Death from any cause	HR 0.99	0.87 to 1.12
Death from cancer	HR 0.83	0.67 to 1.02
Total fractures	HR 0.98	0.89 to 1.08
Hip fractures	HR 1.01	0.70 to 1.47

Nothing crossed into significance except the direction of cancer death, and that interval still touched 1.0.

A 2019 meta-analysis pooled 52 trials covering 75,454 participants and reached the same place with far more statistical power. All-cause mortality: RR 0.98 (0.95 to 1.02), with no heterogeneity between trials at all. Cardiovascular mortality: RR 0.98 (0.88 to 1.08). These are not underpowered nulls or noisy estimates. They are tight intervals centered on no effect.

Fifty-two randomized trials in 75,454 people found no reduction in all-cause mortality from vitamin D supplementation. The confidence interval, 0.95 to 1.02, is narrow enough to rule out the benefit the cohort studies predicted.

The two endpoints that did move

Cancer death. The same meta-analysis found a 16% lower risk of dying from cancer: RR 0.84 (0.74 to 0.95), again with no heterogeneity across trials. VITAL alone produced almost the identical point estimate (HR 0.83) without reaching significance on its own. Cancer death was a secondary endpoint in both, which is a reason for caution rather than dismissal.

Autoimmune disease. A separately registered VITAL ancillary study followed confirmed autoimmune diagnoses and found 123 cases on vitamin D against 155 on placebo: HR 0.78 (0.61 to 0.99), a 22% reduction that cleared significance by the narrowest possible margin.

Two signals from a program that produced nulls everywhere else. Worth watching, not worth acting on yet.

The subgroup everyone assumes was missed

The standard defense of vitamin D runs like this: the trials enrolled well-nourished volunteers who did not need supplements, so of course nothing happened. Give it to genuinely deficient people and you would see something.

VITAL tested that. Among the 15,787 participants with baseline blood samples, mean serum 25(OH)D was 30.8 ng/mL, but 12.7% started below 20 ng/mL and another 32.2% fell between 20 and 30. The 2,001 participants below 20 ng/mL had a cancer hazard ratio of 0.97 (0.68 to 1.39) against 0.98 (0.86 to 1.12) for everyone above. The interaction p-value exceeded 0.50. The fracture ancillary found the same thing: no modification of treatment effect by baseline 25(OH)D level.

The deficient subgroup was smaller and its intervals are wider, so this is not the last word. But the hypothesis was tested rather than ignored, and it did not come out the way its proponents expected.

What these trials did not ask

VITAL enrolled adults who were not selected for deficiency, low bone mass, or osteoporosis. It says nothing about treating diagnosed osteomalacia, rickets, or severe deficiency in someone whose level is genuinely in the teens and who has symptoms to match. Those are clinical situations with their own evidence base, and this page does not evaluate it.

The takeaway

Vitamin D supplementation does not lower all-cause mortality in adults who are not selected for deficiency. That conclusion rests on a 25,871-person randomized trial and a 52-trial meta-analysis that agree with each other, and the intervals are tight enough that a future trial is unlikely to overturn it.

Two endpoints remain open: cancer death, where a 16% reduction showed up in pooled data as a secondary endpoint, and autoimmune disease, where a single ancillary study cleared significance at P = 0.05. Both need confirmation designed to test them directly.

Anyone selling vitamin D as a longevity intervention is selling you the cohort studies and skipping the trials.