Evidence Trace, complete trial record

In plain terms

In plain terms: Vitamin D3 has some human trial evidence supporting it, but the evidence is not yet strong enough for EBL to call it established. It is worth knowing about and worth discussing with your clinician.

Vitamin D3

B Supported: a good trial, backed by more

Category Supplements

Evidence RCTs + meta-analyses (use-specific)

Reviewed July 2026

PMIDs 1331788 · 15886381 · 19797342 · 30415629 · 35939577 · 24922127 · 41571412 · 24729336 · 34656949 · 39993397 · 42161415 · 37380191 · 35676320 · 42704968 · 37011645 · 28384800 · 33170239

The endpoint this grade is based on: Hard clinical endpoint (mortality, major event), Use-specific. Fall prevention in older adults and respiratory infection prevention: Grade C, mixed randomized evidence. Fracture prevention with vitamin D3 alone: Grade D, tested and no benefit, in deficient or older adults and in adults not selected for deficiency. Vitamin D3 with calcium in older adults living in institutions: Grade B, replicated hard-endpoint benefit in institutions, null in the large community trials. Cancer and cardiovascular prevention in adults not selected for deficiency: Grade D, tested and no benefit.

The evidence as the record states it

Evidence record

PMID 1331788 PMID 15886381 PMID 19797342 PMID 30415629 PMID 35939577 PMID 24922127 PMID 41571412 PMID 24729336 PMID 34656949 PMID 39993397 PMID 42161415 PMID 37380191 PMID 35676320 PMID 42704968 PMID 37011645 PMID 28384800 PMID 33170239

Avenell et al. (Cochrane Database Syst Rev, 2014; PMID 24729336) and Massé et al. (BMJ, 2026; PMID 42161415): vitamin D alone does not prevent fractures. Ling et al. (Clin Nutr, 2021; PMID 34656949): no overall effect of vitamin D alone on falls. Jolliffe et al. (Lancet Diabetes Endocrinol, 2025; PMID 39993397): the effect on respiratory infection is no longer significant. Manson et al. (VITAL, 2019; PMID 30415629) and LeBoff et al. (2022; PMID 35939577): no cancer, cardiovascular or fracture benefit in adults not selected for deficiency. Bischoff-Ferrari et al. (JAMA, 2005; PMID 15886381) and Bischoff-Ferrari et al. (BMJ, 2009; PMID 19797342) are earlier positive fracture and fall meta-analyses, superseded by the larger analyses above. Chapuy et al. (1992; PMID 1331788) tested vitamin D3 with calcium, not vitamin D3 alone. Garland et al. (2014; PMID 24922127) is an observational meta-analysis of blood levels and mortality; it carries no grade. Thompson et al. (BMJ, 2023; PMID 37380191), D-Health: 21,315 Australians aged 60 to 84, 60,000 IU monthly for up to five years; major cardiovascular events hazard ratio 0.91 (0.81 to 1.01), myocardial infarction 0.81, a result its authors call consistent with no effect. Keum et al. (Br J Cancer, 2022; PMID 35676320): 12 trials, cancer incidence unchanged (0.99); in six trials, cancer mortality 0.92 overall and 0.87 among trials of daily dosing, a subgroup result. Rikkonen et al. (Bone, 2026; PMID 42704968): 2,495 older Finns, 1,600 or 3,200 IU/day for 5 years; no fracture reduction, and more fractures in the 1,600 IU arm (hazard ratio 1.52). Headline Grade B, use-specific, on vitamin D3 with calcium. Hua et al. (BMJ Open, 2026; PMID 41571412): in 5,065 US adults from the Health and Retirement Study followed for six years, vitamin D supplement users showed slightly faster decline in global cognition (difference -0.052 points per year) than non-users, and only among those whose blood 25(OH)D was already normal at baseline. Observational, so it cannot separate the supplement from the kind of person who takes it; it carries no grade and sits beside the replete-adult D grades as context.

Evidence by specific use

Deficiency correction & fracture risk (deficient or older adults)

Grade D

Tested with vitamin D3 alone, and no benefit. Avenell et al. (Cochrane Database Syst Rev, 2014; PMID 24729336) found high-quality evidence that vitamin D alone does not prevent hip or any fracture; vitamin D with calcium gave a small hip fracture reduction, which is a different intervention and is graded on its own row below. Massé et al. (BMJ, 2026; PMID 42161415), a meta-analysis of 69 randomized trials, found no effect of vitamin D alone on any fracture (RR 1.00). Recent trials in older adults were null: Waterhouse et al. (Lancet Diabetes Endocrinol, 2023; PMID 37011645) and Rikkonen et al. (Bone, 2026; PMID 42704968). The two sources this use previously rested on test something else: Chapuy et al. (N Engl J Med, 1992; PMID 1331788) gave vitamin D3 with calcium to institutionalized women, and the 2005 fracture meta-analysis (PMID 15886381) has been superseded by the larger analyses above. Moved from B in October 2026 by editor adjudication.

PMID 1331788 PMID 15886381 PMID 24729336 PMID 42161415 PMID 37011645 PMID 42704968

Fall prevention (older adults)

Grade C

Mixed. The cited meta-analysis (Bischoff-Ferrari et al., BMJ, 2009; PMID 19797342; 8 trials) was positive, but later and larger evidence is not: Ling et al. (Clin Nutr, 2021; PMID 34656949) found no overall effect of vitamin D alone (RR 1.00), with benefit only in people with low baseline levels. High doses increased falls: Sanders et al. (JAMA, 2010; PMID 20460620) with an annual 500,000 IU dose, and Bischoff-Ferrari et al. (JAMA Intern Med, 2016; PMID 26747333) with monthly 60,000 IU. Appel et al. (Ann Intern Med, 2021; PMID 33284677) found 1,000 IU/day or more did not prevent falls in older adults with low vitamin D and raised fall risk, with safety concerns at higher doses. Two further pooled analyses: Massé et al. (BMJ, 2026; PMID 42161415), 69 trials, found vitamin D alone had no effect on the risk of falling (high certainty), while Wei et al. (Front Endocrinol, 2022; PMID 36034418), 38 trials, found fewer falls at 700 IU a day or more. Pooled analyses disagree: C. Moved from B in October 2026 by editor adjudication.

PMID 19797342 PMID 34656949 PMID 20460620 PMID 26747333 PMID 33284677 PMID 42161415 PMID 36034418

Fracture prevention, vitamin D3 with calcium, older adults living in institutions

Grade B

Mixed, and the split follows the population. Chapuy et al. (N Engl J Med, 1992; PMID 1331788): 3,270 ambulatory women in institutions, mean age 84, 800 IU vitamin D3 with 1.2 g calcium daily for 18 months; hip fractures 43 percent lower and nonvertebral fractures 32 percent lower than double placebo. Avenell et al. (Cochrane Database Syst Rev, 2014; PMID 24729336): nine trials, 49,853 participants, vitamin D plus calcium reduced hip fracture (RR 0.84), which the review translates to nine fewer hip fractures per 1,000 per year in institutions and one fewer per 1,000 in the community. Jackson et al. (N Engl J Med, 2006; PMID 16481635), the Women's Health Initiative: 36,282 community-dwelling postmenopausal women, 400 IU with 1,000 mg calcium for 7 years; hip fracture hazard ratio 0.88 with a confidence interval crossing 1, and renal calculi more frequent. Grant et al. (Lancet, 2005; PMID 15885294), RECORD: 5,292 adults aged 70 or over after a low-trauma fracture, 800 IU with 1,000 mg; no reduction in new fractures. Replicated benefit on a hard endpoint, pooled across nine trials with high-quality evidence by the Cochrane reviewers, and null in the two largest trials outside institutions: B, scoped to older adults living in institutions, the population in which the trials were positive. Not A: the pooled benefit is carried by the institutional trials, and the community trials did not reproduce it. This row grades the combination; vitamin D3 alone for fracture is D Route 2 on the row above.

PMID 1331788 PMID 24729336 PMID 16481635 PMID 15885294

Cancer prevention (already-replete adults)

Grade D

VITAL (Manson et al., N Engl J Med, 2019; PMID 30415629; 25,871 adults not selected for vitamin D status) found no reduction in invasive cancer incidence, and a meta-analysis of 12 trials found none either (Keum et al., Br J Cancer, 2022; PMID 35676320; RR 0.99). Tested; no benefit on incidence. Cancer death is a different endpoint: the same meta-analysis found a lower cancer mortality among trials of daily dosing (0.87), a subgroup result that this row does not grade.

PMID 30415629 PMID 35676320

Cardiovascular prevention (already-replete adults)

Grade D

VITAL (Manson et al., N Engl J Med, 2019; PMID 30415629) found no reduction in major cardiovascular events. D-Health (Thompson et al., BMJ, 2023; PMID 37380191; 21,302 older Australians, 60,000 IU monthly) found a borderline lower rate (hazard ratio 0.91, 0.81 to 1.01) with fewer heart attacks, which its authors describe as consistent with no effect. ViDA (Scragg et al., JAMA Cardiol, 2017; PMID 28384800; 5,110 adults aged 50 to 84 in New Zealand, 200,000 IU then 100,000 IU monthly for a median 3.3 years) found no reduction in cardiovascular disease (hazard ratio 1.02, 0.87 to 1.20), including in the quarter who were deficient at baseline. Three large trials without a clear benefit: tested, no benefit.

PMID 30415629 PMID 37380191 PMID 28384800

Fracture prevention (already-replete adults)

Grade D

VITAL ancillary (LeBoff et al., N Engl J Med, 2022; PMID 35939577) found supplemental vitamin D did not reduce fractures in generally healthy midlife and older adults not selected for vitamin D status. D-Health (Waterhouse et al., 2023; PMID 37011645) and the Finnish FIND trial (Rikkonen et al., Bone, 2026; PMID 42704968) were also null, and FIND recorded more fractures in its 1,600 IU/day arm (hazard ratio 1.52) while 3,200 IU/day did not differ from placebo. DO-HEALTH (Bischoff-Ferrari et al., JAMA, 2020; PMID 33170239; 2,157 adults aged 70 and over, 2,000 IU/day for 3 years) found no effect on nonvertebral fractures. Tested; no benefit. Vitamin D3 alone in deficient or older adults is also D on the row above; only vitamin D3 with calcium in institutions carries the B.

PMID 35939577 PMID 37011645 PMID 42704968 PMID 33170239

Prevention of acute respiratory tract infection (esp. in vitamin D-deficient people)

Grade C

Mixed. Martineau et al. (BMJ, 2017; PMID 28202713) pooled individual-patient data from randomized trials and found a modest reduction, largest in very deficient people. The same group's update, Jolliffe et al. (Lancet Diabetes Endocrinol, 2025; PMID 39993397), with 40 trials, found the overall effect no longer statistically significant (OR 0.94, 95% CI 0.88 to 1.00) and no modification by baseline vitamin D status. A test-and-treat trial in people with low levels, Jolliffe et al. (BMJ, 2022; PMID 36215226), found no reduction in acute respiratory infection. Moved from B in October 2026 by editor adjudication.

PMID 28202713 PMID 39993397 PMID 36215226

All-cause mortality (older adults not selected for vitamin D status)

Grade D

D-Health (Neale et al., Lancet Diabetes Endocrinol, 2022; PMID 35026158): 21,315 Australians aged 60 and over, 60,000 IU monthly for 5 years; deaths 5.3 against 5.1 percent (hazard ratio 1.04, 0.93 to 1.18), with point estimates the authors describe as consistent with more cancer deaths. Zhang et al. (BMJ, 2019; PMID 31405892), 52 randomized trials and 75,454 participants: no effect on all-cause mortality (risk ratio 0.98, 0.95 to 1.02). Cancer death was lower in that pool (0.84), a different endpoint this row does not grade. A large trial built for this endpoint and a large pooled analysis, both null: tested, no benefit.

PMID 35026158 PMID 31405892

Progression from prediabetes to type 2 diabetes

Grade C

D2d (Pittas et al., N Engl J Med, 2019; PMID 31173679): 2,423 adults with prediabetes not selected for vitamin D status, 4,000 IU a day for a median 2.5 years; diabetes in 9.39 against 10.66 per 100 person-years (hazard ratio 0.88, 0.75 to 1.04), not significant. The same research group's pooled analysis of three trials (Pittas et al., Ann Intern Med, 2023; PMID 36745886), which includes D2d and a trial of eldecalcitol, a prescription vitamin D analog, found 15 percent fewer cases (hazard ratio 0.85, 0.75 to 0.96). The largest trial alone was null, and the positive pool rests on it and on a drug that is not vitamin D3: mixed, C. New-onset diabetes here is a diagnosis defined by glucose thresholds.

PMID 31173679 PMID 36745886

Prevention of depression in adults aged 50 and over

Grade D

VITAL-DEP (Okereke et al., JAMA, 2020; PMID 32749491): 18,353 adults aged 50 and over, 2,000 IU a day for a median 5.3 years; depression or clinically relevant symptoms in 12.9 against 13.3 per 1,000 person-years (hazard ratio 0.97, 0.87 to 1.09), with no difference in first episodes, recurrences or mood scores (mean difference 0.01 points). One large trial built for this endpoint, null: tested, no benefit, on the same footing as other single-trial Route 2 rows. This does not grade vitamin D as a treatment for people who already have depression.

PMID 32749491

Why Grade B and not Grade A

Grade B reflects moderate human evidence -- typically a single strong trial, multiple smaller trials, or surrogate endpoint data without hard clinical outcome replication. To reach Grade A, Vitamin D3 would require at least one additional independent randomized trial demonstrating a hard clinical endpoint or objectively measured functional outcome in a comparable population.

What this grade covers and what it does not

Headline Grade B. The headline follows the best-supported use: vitamin D3 with calcium for fracture prevention in older adults living in institutions, B on replicated hard-endpoint benefit in institutions with the large community trials null. Fracture prevention with vitamin D3 alone moved from B to D on Route 2 in October 2026, because the Cochrane review and a 69-trial meta-analysis found no effect of vitamin D alone and the trial previously cited gave calcium as well. Falls and respiratory infection are C on mixed randomized evidence. Cancer, cardiovascular and fracture prevention in adults not selected for vitamin D status are D on Route 2. WHAT THIS GRADE DOES NOT COVER: treating diagnosed vitamin D deficiency or conditions such as rickets and osteomalacia, which are medical treatment rather than supplementation; and vitamin D3 on its own for fracture, which is D. Safety signals from the trials: annual 500,000 IU and monthly 60,000 IU doses increased falls, and the 1,600 IU/day arm of the Finnish trial had more fractures than placebo.

What would move this grade

An independent randomized trial demonstrating a hard clinical endpoint -- mortality, major adverse event, or objectively confirmed functional improvement -- in an adequately powered sample, with results independently replicated, would support Grade A.

How to track whether this is working 25(OH)D blood level, for which no target is established and which is read against the clinical indication. Retested at 12 weeks after starting or changing dose, then every 6 months until stable. Calcium and PTH where dosing exceeds 5,000 IU/day.