Across 46 trials, vitamin D pills didn't cut fractures or falls

A review of 69 trials and 153,902 adults found that 6.5% of people broke a bone whether they took vitamin D or a dummy pill. Calcium and vitamin D together did slightly better, about 1 fewer fracture per 100 people, mostly thanks to one 1992 trial in French care homes. Most volunteers lived at home and weren't short of vitamin D; people who are truly deficient were barely tested.

5 October 2026 · How we rate evidence

In 1992, a trial in French care homes gave 1,634 women, average age 84, calcium and vitamin D every day, and gave 1,636 others dummy pills. Over three years, about 16 in 100 of the supplemented women broke a bone outside the spine, against 19 in 100 of the others. Many of these women had very little vitamin D in their blood and got little calcium from food. That result helped launch decades of advice to take both. More than 30 years and dozens of trials later, a review in The BMJ finds that the 1992 trial is still doing most of the work.

The claim going around. The review came out on 20 May 2026 with a BMJ press release saying the supplements "offer little to no clinically meaningful benefit on fracture and fall prevention in most older people." ScienceDaily reposted that release on 20 September as "Millions take calcium and vitamin D for bone health. A major review finds little benefit," and a second wave followed, including Labroots ("Calcium and Vitamin D May Not Prevent Fractures") and The Economic Times. On the whole, the coverage describes the paper fairly.

What they did. A Montreal pharmacy team led by Olivier Massé gathered every randomised trial they could find that compared calcium, vitamin D, or both with a dummy pill or nothing, in adults not taking osteoporosis drugs. Randomised means a coin toss, not the patient, decided who got the supplement. That left 69 trials and 153,902 people. They pooled the results for each supplement separately. The main question, registered before they started, was whether fewer people had a fracture of any kind; falls and specific fractures were secondary questions.

The paper says they also fixed in advance what they would count as worth having: 2 fewer people per 100 with any fracture, 0.7 per 100 with a hip fracture, 3 per 100 falling. The authors set them by consensus and call them arbitrary themselves, and we couldn't find them in the main text of the published protocol.

What they found. Vitamin D on its own did nothing measurable. In 36 trials with 92,415 people, 6.5% had a fracture whether they took vitamin D or not. The range around that result rules out a benefit bigger than about 3 fewer fractures per 1,000 people, far below the 20 per 1,000 the authors set as worthwhile. Falls told the same story: in 32 trials with 65,234 people, a quarter fell in both groups. Hip fractures were, if anything, slightly more common with vitamin D (1.4% against 1.3%), a difference at the edge of chance.

Calcium alone went with 1 fewer person in 100 having a fracture (9.9% against 10.9%), but the range still allows no effect at all. Calcium plus vitamin D is where the review found something: 10.3% of people had a fracture against 11.3% without, and hip fractures fell from 1.8% to 1.5%. Put in absolute terms, that is 1 fracture prevented for every 100 people taking the pills for about two years, and 3 hip fractures per 1,000.

Most of that comes from the 1992 trial. The largest trial in the pool, the US Women's Health Initiative with 36,282 women, found 11.6% against 11.9%, a difference not clear of chance. Take out the French trial (we identified it from the forest plot) and the overall result shrinks to a relative 4% and loses its statistical clarity. So does keeping only trials at low risk of bias (0.94), or only trials that counted every kind of fracture (0.93). The paper reports all of these checks itself. Because the 1.0-point gap fell below their 2-point threshold, the authors rate it "high certainty of little to no effect".

What it doesn't show. It can't say much about the people most likely to benefit. Only 2 of 69 trials enrolled people whose average vitamin D was below 25 nmol/L, the level often used for deficiency. The authors' test of vitamin D in trials starting below 50 nmol/L found nothing (7,509 people), but the only trial below 25 had 105 people. For calcium plus vitamin D, too few trials reported starting levels to test at all.

Care homes are the other gap. In the three calcium-plus-vitamin D trials run there, with 4,014 residents, fractures were 15% lower, against 4% lower in community trials. Both are relative figures; the paper gives no absolute rates for these groups. But by our calculation about 85% of that result's weight is the same 1992 trial, so it isn't a separate signal. Of the other two trials, one (583 women) found no difference and one had just 6 fractures. The review's "age 80 or over" group is the same three trials. And the test of whether care homes really differ from the community wasn't clear of chance. The authors say their absolute figures "may somewhat underestimate benefits" in such groups.

There's also a bias toward finding nothing, which the authors acknowledge: almost all trials let control groups take their own supplements, or didn't say they couldn't. And the verdict of "little to no" for calcium plus vitamin D depends on thresholds the authors chose. Someone who thinks 3 fewer hip fractures per 1,000 is worth a daily pill could read the same numbers differently.

What it means beyond the study. For people, the question is whether a daily supplement protects bones. For most healthy older adults living at home, these trials give a fairly firm answer for vitamin D alone: no. The review doesn't cover people on osteoporosis drugs, where calcium and vitamin D are given as part of the treatment. Nor can it tell someone with a measured deficiency, or a frail care-home resident, that the pills are pointless. For care facilities, a 2025 Cochrane review, which probably shares some trials with this one, reports that vitamin D went with fewer falls in total among residents with low levels, though not with fewer residents falling.

Why it's still interesting. Settling a question with "no" is hard, and this is what it looks like when it's done well. Tens of thousands of people and a range narrow enough to rule out anything worth having. It also shows how one striking early trial in an unusual group can shape advice for everyone for decades. The 1992 women were very old, lived in care homes and were badly short of vitamin D. Most people buying supplements today are none of those.

Open questions.

Do people who are truly deficient benefit? Almost no trial tested only people with very low vitamin D. A trial that screens first and randomises only those below about 25 nmol/L, large enough to count fractures, would answer it. If they benefit, testing and treating a few makes sense even though treating everyone doesn't.

Is the care-home signal real? It rests mostly on the 1992 trial; the two smaller care-home trials added little. The gap from community trials could be chance. A new placebo-controlled trial in care homes, or a pooled analysis of individual residents' data from existing trials, would settle whether routine supplements there should stay or go.

How big a benefit is worth a daily pill? The review's thresholds came from the authors' consensus, not from patients. Asking older adults how many fractures prevented would justify years of pills would show whether calcium plus vitamin D's small effect counts as nothing or as something.

A note on timing. ScienceDaily's repost came four months after the review's 20 May 2026 publication; the evidence hadn't changed.