This is one of the few nutrition questions where the answer is not in dispute between researchers but between researchers and the shelf. Vitamin D supplementation reduces fractures: this is what a large part of the public believes. Here is what the randomized record actually contains, in the order the numbers were printed.
The biggest trial, read carefully
VITAL is a two-by-two factorial randomized controlled trial of vitamin D3 at 2,000 IU per day and n-3 fatty acids at one gram per day, in US men aged 50 or older and women aged 55 or older. The fracture paper is an ancillary study of it. The design point comes first: participants were not recruited on the basis of vitamin D deficiency, low bone mass or osteoporosis.
25,871 participants. 50.6 per cent women (13,085 of 25,871), 20.2 per cent Black (5,106 of 25,304). 1,991 incident fractures in 1,551 participants over a median follow-up of 5.3 years, reported on annual questionnaires and adjudicated by centralised medical-record review.
Total fractures: 769 of 12,927 in the vitamin D group against 782 of 12,944 in the placebo group. Hazard ratio 0.98 (95 % CI 0.89 to 1.08), p = 0.70. Nonvertebral fractures, hazard ratio 0.97 (95 % CI 0.87 to 1.07), p = 0.50. Hip fractures, hazard ratio 1.01 (95 % CI 0.70 to 1.47), p = 0.96.
And the subgroup sentence that closes off the usual escape route: no modification of the treatment effect according to baseline characteristics, including age, sex, race or ethnic group, body-mass index, or serum 25-hydroxyvitamin D levels. There was no subgroup in which vitamin D did something different.
Thirty-three trials pooled
The 2017 JAMA meta-analysis did something VITAL could not: it pooled 33 randomized trials with 51,145 participants, comparing calcium, vitamin D, or both against placebo or no treatment for fracture incidence in community-dwelling adults over 50.
Hip fracture with calcium: risk ratio 1.53 (95 % CI 0.97 to 2.42), absolute risk difference 0.01 (95 % CI 0.00 to 0.01). With vitamin D: risk ratio 1.21 (95 % CI 0.99 to 1.47). With combined calcium and vitamin D: risk ratio 1.09 (95 % CI 0.85 to 1.39). Nothing significant, and nothing significant for nonvertebral, vertebral or total fractures either.
The subgroup analysis is the part worth reading twice: these results were generally consistent regardless of the calcium or vitamin D amount, sex, fracture history, dietary calcium intake, or baseline serum 25-hydroxyvitamin D concentration. The conclusion the authors draw is that the findings do not support the routine use of these supplements in community-dwelling older people.
The risk ratios above 1.0 are what happens when a null result is displayed honestly rather than as a protective effect. Nobody should read 1.21 as vitamin D causing hip fractures either — the interval crosses 1.
The one large trial that did report a renal signal
The Women’s Health Initiative calcium-and-vitamin-D trial randomised 36,282 postmenopausal women aged 50 to 79 to one gram of elemental calcium as calcium carbonate with 400 IU of vitamin D3 daily, or to placebo, with fractures ascertained over an average of 7.0 years.
Hip bone density was 1.06 per cent higher in the treatment group (p < 0.01). Intention-to-treat hazard ratios: 0.88 for hip fracture (95 % CI 0.72 to 1.08), 0.90 for clinical spine fracture (95 % CI 0.74 to 1.10), 0.96 for total fractures (95 % CI 0.91 to 1.02). And renal calculi: hazard ratio 1.17 (95 % CI 1.02 to 1.34).
So in the largest fracture trial of all, the density endpoint moved significantly, the fracture endpoints did not, and the kidney-stone endpoint moved the other way. Effects did not vary significantly by prerandomisation serum vitamin D levels.
One number in that abstract deserves its own paragraph, because it is the kind of thing that gets quietly dropped. When women were censored at the point they stopped adhering to the study medication, the hazard ratio for hip fracture became 0.71 (95 % CI 0.52 to 0.97). The authors report it and the framing is neutral, but it is a compliance-adjusted estimate in a trial whose primary analysis was null, and reading it as “vitamin D works if you take it” is exactly the substitution this blog exists to catch.
What the smaller trials measured instead
A 2023 systematic review in Nutrition of vitamin D supplementation in postmenopausal women without osteoporosis identified 3,460 studies and selected nine. Its endpoints are the biomarkers.
Vitamin D supplementation raised 25-hydroxyvitamin D levels and reduced parathyroid hormone secretion, dependent on baseline levels. 400 IU improved the percentage of carboxylated osteocalcin. 800 to 1,000 IU combined with calcium gave reduced, improved or maintained bone mineral density and reduced alkaline phosphatase. 4,000 IU alone or with calcium for six months did not improve C-telopeptide or procollagen type 1 peptide levels. 15,000 IU per week increased cortical area of metacarpal bone. 500,000 IU annually for five years did not contribute to reducing fracture risk and falls.
Only one of the nine found a reduction in fracture risk: 800 IU of vitamin D with 1,200 milligrams of calcium.
This is the shape of the whole field. Surrogate endpoints move, quite consistently, and the clinical endpoint does not follow — the authors’ own conclusion is that it was not possible to assert that supplementation reduced fracture risk in postmenopausal women.
Who is actually taking it
A 2025 cross-sectional analysis of the OPERAM trial in four European countries looked at 2,008 multimorbid older patients — mean age 79.4 years, 45 per cent female, all on five or more chronic medications. 825 of 2,008 (41.1 per cent) were supplemented with vitamin D.
Then the finding that argues for reading labels rather than headlines: 681 participants (33.9 per cent of the whole cohort, 69.7 per cent of non-users) had potential underuse by the START criteria, and 204 (10.2 per cent of the cohort, 24.7 per cent of users) had potential overuse. Increasing age and being male were associated with underuse. The number of baseline medications moved both: underuse decreased with each additional medication (odds ratio 0.93, 95 % CI 0.90 to 0.95) and overuse increased (odds ratio 1.08, 95 % CI 1.04 to 1.12).
If a supplement is taken by 41 per cent of the frailest, most medicated people in Europe, and one in ten of them may be taking it when the criteria say they should not, that is a supply-side story as much as a demand-side one.
How to read the claim
Four questions. Is the endpoint a fracture or a marker — most of this literature is about density, carboxylated osteocalcin, alkaline phosphatase, and parathyroid hormone, and only two of the five papers here report fractures as an outcome. Was the population selected for deficiency, as WHI and VITAL explicitly were not. Was the trial powered for hip fracture specifically, or for total fractures, where events dilute any effect. And is the quoted figure the intention-to-treat one or the adherent one.
What this evidence does not show
It does not show that vitamin D is useless: serum deficiency is real, and none of these trials enrolled people by their vitamin D level alone. It does not show that supplementation has no effect in people who are deficient — these trials mostly did not test that population, which is a different claim from the one this article examined. It does not address falls, muscle outcomes or seasonal respiratory infection, where the literature is a separate body. And a hazard ratio of 1.17 for kidney stones in one trial of a combined calcium and vitamin D product is not a statement about vitamin D alone.
This article summarises published randomized trials and systematic reviews. It is not medical advice and it is not a recommendation for any individual.
