Vitamin D supplementation: evidence and dosing

An evidence-grade summary of the trial-level vitamin D literature, including VITAL, what dosing is supportable, and where the evidence stops.

Background

Vitamin D is unique among the fat-soluble vitamins because endogenous synthesis from skin exposure to UVB radiation, not dietary intake, is the dominant input in most populations. Latitude, season, skin pigmentation, age, body composition, and time-outdoors patterns all affect endogenous synthesis, which is why a substantial fraction of adults in temperate latitudes maintain marginal serum 25-hydroxyvitamin D (25(OH)D) concentrations for at least half the year despite an apparently adequate dietary intake. Supplementation is the most reliable way to raise serum levels in individuals with low endogenous synthesis.

The clinical question is not whether vitamin D supplementation raises serum 25(OH)D — it does, predictably, in a dose-dependent manner — but whether raising serum 25(OH)D in apparently healthy adults improves clinical outcomes. The trial-grade evidence on this question is now substantial, and the answer is more nuanced than the popular discussion of vitamin D often allows.

Skeletal outcomes: where the evidence is strongest

The mechanistic case for vitamin D in bone health is unambiguous: vitamin D is required for active intestinal calcium absorption, and severe deficiency produces rickets in children and osteomalacia in adults. Trial-grade evidence on fracture prevention in older adults shows that combined vitamin D and calcium supplementation reduces fracture risk in institutionalized populations and in vitamin-D-deficient community-dwelling older adults, with a smaller and less consistent effect in replete populations (Avenell et al., 2014, Cochrane).

Bischoff-Ferrari and colleagues’ DO-HEALTH trial (2020) tested 2,000 IU/day vitamin D3, omega-3 fatty acids, and a home strength-training program in 2,157 generally healthy adults aged 70 and over. None of the three interventions, alone or in combination, produced a clinically meaningful improvement in the primary composite outcome of physical function, fracture incidence, blood pressure, infection rate, or cognitive function over three years. The trial supports the position that incremental vitamin D supplementation in already-replete older adults does not produce additional skeletal or non-skeletal benefit at the tested dose.

VITAL and the non-skeletal endpoints

VITAL (Manson et al., 2019) is the largest randomized trial of vitamin D supplementation conducted to date. It enrolled 25,871 US adults (men aged 50 and over, women aged 55 and over) and randomized them in a 2x2 factorial design to 2,000 IU/day vitamin D3 or placebo and to 1 g/day marine omega-3 or placebo. Median follow-up was 5.3 years. Mean baseline serum 25(OH)D was 30.8 ng/mL — a generally replete population.

The primary results were null: vitamin D3 did not reduce the primary cardiovascular composite (HR 0.97, 95% CI 0.85 to 1.12) or the primary cancer composite (invasive cancer of any type; HR 0.96, 95% CI 0.88 to 1.06). Secondary analyses produced more interesting signals — a possible reduction in cancer mortality after exclusion of the first two years of follow-up, and a reduction in incident autoimmune disease in a separate VITAL substudy — but these are exploratory and would require confirmation in trials designed to test them as primary endpoints.

The VITAL design has two important interpretive constraints. The trial population was largely replete at baseline, so the result speaks to the effect of additional supplementation in adults who were not deficient, not to the effect of correcting deficiency. The dose tested was 2,000 IU/day; the trial does not directly address the higher doses (4,000 to 10,000 IU/day) that some clinicians prescribe for documented insufficiency.

Dosing in practice

The Institute of Medicine (2011) Dietary Reference Intakes set the RDA for adults at 600 IU/day (15 mcg/d) for ages 19 to 70 and 800 IU/day for adults over 70, with a tolerable upper intake level (UL) of 4,000 IU/day. The Endocrine Society guideline (Holick et al., 2011) recommends higher therapeutic doses (1,500 to 2,000 IU/day for general maintenance, 6,000 IU/day temporarily for treatment of documented deficiency) and uses a higher serum sufficiency threshold (30 ng/mL versus the IOM’s 20 ng/mL). The two guidelines disagree because they apply different definitions of “sufficiency” and weight the trial evidence differently; both positions are defensible within the published data.

Practical positions that the data support:

• For an adult at risk of inadequacy (low sun exposure, higher BMI, darker skin pigmentation at higher latitudes, advanced age, malabsorption): 800 to 2,000 IU/day vitamin D3 is supportable as a maintenance dose. Vitamin D2 is approximately equivalent over chronic dosing although less efficient on a single-dose basis.
• For an adult with measured serum 25(OH)D below 20 ng/mL: short-term higher doses (commonly 4,000 to 6,000 IU/day for 8 to 12 weeks) followed by maintenance dosing and re-measurement is the conventional approach and is supported by Endocrine Society guidance.
• For an adult with serum 25(OH)D above 30 ng/mL and no specific clinical indication: the marginal benefit of additional supplementation is small and the VITAL data argue against expecting cardiovascular or cancer outcome benefit from incremental dosing.

Doses above the IOM upper limit of 4,000 IU/day are not supported by hard-endpoint efficacy data outside specific clinical indications and have been associated with hypercalcemia at sustained higher daily intakes. The relevant individualization tool is serum 25(OH)D measurement, not symptom self-report; the symptoms attributed to mild vitamin D inadequacy (fatigue, mood, generalized aches) are non-specific and the placebo response in supplementation trials is substantial.

Limitations of the evidence base

The vitamin D trial literature has two recurring methodological limitations. First, baseline serum 25(OH)D varies widely across trial populations, and trials in already-replete populations are unlikely to detect benefit even if a real effect exists at lower starting concentrations. Second, the relationship between serum 25(OH)D and clinical outcomes appears non-linear, with diminishing returns above approximately 30 ng/mL — meaning that a fixed-dose trial in a heterogeneous population is the wrong design for detecting an effect that depends on the deficiency-correction increment. Future trials targeting deficient populations specifically, with serum 25(OH)D as a screening criterion, are likely to be more informative than the next large general-population trial (Bouillon et al., 2019).