Vitamin D
cholecalciferol

Its best-demonstrated role is in bones and muscles: with sufficient vitamin D levels, the body absorbs dietary calcium and bone stays strong. In older people or those with a deficiency, correcting it helps bone and muscle health.

There is also solid evidence on one point: when someone starts from a significant deficiency, correcting it is associated with fewer respiratory infections. The benefit lies in filling the gap, not in stockpiling more.

What vitamin D does not do, according to the large trials, is as important as what it does. In people who already have normal levels, taking supplements has not been shown to prevent cancer, heart attacks, fractures, falls, depression or diabetes. It sounds like a brochure promise, but the large, well-conducted studies do not support it.

It is worth pausing on why the list of “does not do” is so long, because it is not an oversight by science: it is its finding. For years, observational studies saw that people with low vitamin D got sicker from almost everything, and it was inferred that raising it would protect them. When those ideas were finally put to the test with large, well-designed trials, the protection vanished. VITAL, with nearly 26,000 people, found no reduction in cancer or heart attacks. D-Health, with more than 21,000, found no reduction in mortality. VITAL-DEP found no reduction in depression. D2d found no reduction in diabetes except in a small group starting from profound deficiency. The details of each are further down, in the study cards.

It makes most sense for those with a real chance of running low: older people, especially if they rarely leave the house; people with dark skin living at high latitudes; people with obesity; those who barely get any sun or always stay well covered; and those with intestinal absorption problems.

Deficiency is more common than intuition suggests. In Europe, with standardized data from nearly 56,000 people, 13% have frankly low levels (below 30 nmol/L) and 40% do not reach the sufficiency threshold (50 nmol/L). In winter the figure worsens. This is not a problem of gray, faraway countries: it cuts across all of them.

Here is the fact that throws almost everyone off: southern Europe has more vitamin D deficiency than the north. The Nordic countries hover below 20% deficiency; those of southern and eastern Europe, between 30% and 60%. Spain fits that uncomfortable pattern. In a Catalan population study with more than half a million people, over 75% had suboptimal levels, and the figure climbed toward 80% among young people.

The maintenance doses commonly used range between 800 and 2,000 IU a day. European and U.S. agencies set 4,000 IU/day as the maximum level they consider free of appreciable risk for adults; above that figure, the decision belongs to a professional.

One warning that does matter: more is not better. Very widely spaced megadoses (those annual or monthly ampoules taken all at once) not only offer no advantage but, in a trial with a very high annual dose, increased falls and fractures. The regimen with the best profile is the daily one (or weekly), in moderate amounts.

On the fashionable cofactors we have to be honest. You often read that vitamin D needs magnesium to be activated, or that combining it with vitamin K2 enhances its effects on bone. Magnesium does take part as a cofactor in its metabolism, and the K2 hypothesis is biologically reasonable, but there are no solid trials showing that adding either one to vitamin D improves any clinical outcome in people who are not deficient. It is an evidence gap, not an established recommendation.

Within the usual maintenance doses (up to 4,000 IU/day in adults), vitamin D has a wide safety margin. Toxicity is rare and appears with very high intakes sustained over a long time, not with a normal dose.

It is worth spelling out why those situations call for supervision, because this is not generic caution. There are conditions in which the body produces calcitriol on its own or handles calcium poorly (sarcoidosis and other granulomatous diseases, primary hyperparathyroidism, some kidney stones and certain forms of renal insufficiency) and in them adding vitamin D can throw blood calcium out of balance. In fact, several of these conditions were exclusion criteria in the large trials.

This guide is written for the adult considering taking vitamin D on their own, and there the main story is deficiency versus repletion. In pregnancy, breastfeeding and childhood the ground is different and deserves an honest mention: correcting deficiency does matter (it prevents childhood rickets and there are signals, still inconclusive, about a possible effect on infant wheezing), but the doses, the thresholds and the decisions are the responsibility of the pediatrician or obstetrician, not of an informational guide.

We therefore leave it as a declared gap: it is a subject with an identity of its own that would merit its own guide, and of which we note here only that it exists and that it is decided in consultation, not by reading labels. The safe upper limit for children aged 1 to 10 is lower than the adult one (2,000 IU/day), one more reason not to improvise.

The idea that vitamin D protects the brain is attractive and has a biological basis (there are receptors on the neuron), but trials do not support it in a healthy population. The Finnish trial designed to see whether it prevented dementia found no differences, and the cognitive measures of VITAL and DO-HEALTH found none either. As in the rest of this guide, meta-analyses leave a crack open only for vulnerable or deficient people, not for those already replete. On mood, the large specific trial (VITAL-DEP) did not find that it prevented depression.

Vitamin D’s promises have been put to the test in some of the largest trials that exist for a supplement. These cards summarize the ones that weigh most: the one that established “it does not prevent disease in replete people”, the one that left the only solid positive signal, the one that marked the safety limit and the context on respiratory infections.

Trials have closed the question “should everyone supplement?” with a fairly firm no, but they have left others open that are worth presenting for what they are: doubts, not controversies.

What if the low level is a consequence, not a cause? The deepest objection to all observational vitamin D research. That those who fall ill have low 25(OH)D may be because the illness, a sedentary life, obesity or spending little time outdoors lowers vitamin D, and not the other way around. Mendelian randomization studies, which get around that problem, mostly find no causal relationship with mortality, except perhaps in the most extreme deficiency. If that is so, a good part of vitamin D’s “promise” was a statistical mirage.

The problem of recruiting those who are already fine. VITAL, D2d and DO-HEALTH mostly included people who were already sufficient. A trial cannot correct a deficiency that does not exist, so their null results might reflect the design rather than the absence of a biological effect in those who are genuinely deficient. That is why “it does not work” and “it has not been properly tested where it would work” coexist in the literature. The targeted hypothesis (supplementing only the deficient) still lacks a large trial to settle it.

Signals pointing in opposite directions. Cancer mortality illustrates how slippery the ground is. In VITAL, with a daily dose, a possible reduction appeared when the first years were excluded; in D-Health, with a monthly dose, the trend was the opposite, toward more cancer mortality (HR 1.24 excluding two years; p=0.05). It may be an effect of the regimen (daily versus bolus), it may be chance. Neither signal was the primary objective of its trial, and neither is enough to assert anything.

“Sufficient” does not mean the same for everyone. Agencies do not agree on the threshold. The U.S. reference bodies set bone sufficiency at 20 ng/mL; the Endocrine Society argued in 2011 for reaching 30, but in 2024 it abandoned that figure and stopped endorsing “optimal” thresholds in healthy people, acknowledging that the evidence does not show a clear net benefit above sufficiency. When the very concept of “being low” changes depending on who defines it, any headline about “X% of the population being deficient” has to be read while checking which threshold was used.