The strongest single result is on autoimmune disease — a five-year trial cut new cases of rheumatoid arthritis, psoriasis, autoimmune thyroid disease, and polymyalgia by 22% in people taking 2,000 IU a day. Add a meaningful drop in mortality risk for the deficient, fewer respiratory infections through winter, fewer falls and fractures in older adults at the right dose. The catch: those benefits land on the third of the population that's actually deficient. If you're already at a healthy serum level — outdoor work, pale skin, low latitude — the gain is essentially zero. Cost is trivial, downside is essentially zero below 4,000 IU/day, and the right question isn't should I take it? but am I in the group that benefits?
Vitamin D isn't really a vitamin. The body makes it on demand from cholesterol in skin cells, but only when sunlight in a narrow ultraviolet band — UVB, around 290 to 315 nanometres — actually reaches the skin. The further you live from the equator, the more atmosphere those photons have to cross in winter, and the more of them are absorbed before they get to you. Above roughly the 40th parallel, the winter sun is so flat that almost no UVB makes it through. Your skin can sit in bright noon sun in January and produce nothing.
Skin colour is the other axis. Melanin absorbs the same wavelengths the body needs for the chemistry — that's most of what melanin evolved to do. Darkly pigmented skin needs roughly three to five times the sun exposure of lightly pigmented skin to produce the same amount Holick 2007. The lighter complexions that appeared in human populations as they moved north weren't aesthetic; they were a vitamin-D solution. A person with deep brown skin living in northern England has the latitude problem and the pigmentation problem stacked, and modern indoor life takes care of whatever sun exposure was left.
Whatever the skin makes — or whatever you swallow — gets converted in the liver into the storage form (25-hydroxyvitamin D, the number a blood test measures), and then in the kidney into the active hormone (1,25-dihydroxyvitamin D). That active form binds a receptor expressed in nearly every cell type in the body. The classical job is calcium and bone. The receptor's presence in immune cells, muscle, and brain is why deficiency leaks into immunity, fall risk, autoimmune disease, and possibly mortality.
What the trials actually show
The headlines you've seen — "biggest vitamin D trial finds no benefit" — are mostly real, and mostly being misread. Five of the largest randomised trials of the last decade tested supplementation against placebo, on hard endpoints, in tens of thousands of people. Most were null. The interpretation hinges on a detail that gets lost in the coverage: the trials enrolled adults who were already, on average, fine.
The trials that recruited people who were genuinely low at baseline tell a different story. Bischoff-Ferrari's pooled analysis of fracture trials found that 700 to 800 IU a day cut hip fractures by about a quarter and any non-vertebral fracture by a similar margin — but 400 IU a day didn't Bischoff-Ferrari et al. 2005. Her fall-prevention meta-analysis, same dose threshold, cut falls in older adults by about a fifth Bischoff-Ferrari et al. 2009. Martineau's individual-participant meta-analysis of 25 trials of respiratory infection found an overall 12% reduction in colds, flu, and pneumonia in adults taking daily or weekly D — but in the subgroup that started deeply deficient (below 25 nmol/L), the reduction was 70% Martineau et al. 2017. That's not a marginal trial result; that's one of the largest preventive-medicine effects in the modern literature, in the subgroup where the intervention was correcting an actual shortage.
And then there's the autoimmune signal. The VITAL ancillary on incident autoimmune disease — rheumatoid arthritis, polymyalgia rheumatica, autoimmune thyroid disease, psoriasis — found a 22% reduction in confirmed new cases over five years on 2,000 IU/day Hahn et al. 2022. The cohort wasn't selected for deficiency, the endpoint was adjudicated, and the magnitude is hard to ignore.
The shape that emerges across the whole literature: vitamin D corrects deficiency. It does not act as a tonic on people who aren't deficient. The big null trials and the positive subgroups aren't contradicting each other — they're describing two different populations.
Who is actually deficient
About a third of US adults sit below the 50 nmol/L (20 ng/mL) threshold the Institute of Medicine considers sufficient, and roughly seven in ten fall below the 75 nmol/L line older guidelines used to draw. Worldwide, a pooled analysis of 7.9 million people found similar numbers Cui et al. 2023. The risk isn't evenly distributed; it concentrates in clear groups:
- Living far from the equator. Above about 40°N (a line that runs through New York, Madrid, Beijing, Istanbul) or below 40°S, the winter sun doesn't generate any usable vitamin D from late autumn to early spring. Above 50°N (London, Berlin, Vancouver, Moscow) the off-season stretches six months.
- Darker skin. Higher melanin needs three to five times the exposure for the same yield Holick 2007. African Americans show the highest deficiency rates in US surveys.
- Age 75 and over. Older skin makes less of the precursor; older kidneys activate less of what gets made. The 2024 Endocrine Society guideline singles this group out explicitly: supplement empirically, no need to test first Demay et al. 2024.
- Indoor lives. Office workers, night-shift workers, anyone who covers most of their skin for cultural or religious reasons, anyone in a long-term-care setting.
- Obesity. Vitamin D is fat-soluble and gets sequestered in adipose tissue. A BMI above 30 means lower serum levels for the same dose.
- Pregnancy. Requirements rise; deficiency is linked to pre-eclampsia and adverse infant outcomes. The 2024 guideline recommends supplementation throughout Demay et al. 2024.
- Malabsorption. Crohn's, celiac, bariatric surgery, cystic fibrosis. Higher doses required; talk to a clinician.
The reader who probably doesn't need to think about this: lightly pigmented skin, lives below 35°N (the latitude of Atlanta, Tokyo, Sydney), spends time outdoors most days, not over 75, not pregnant, not obese. Most other readers benefit from a default of 1,000 to 2,000 IU/day, year-round.
What deficiency actually costs you
At the severe end — a number you'd see in a homebound older adult, or a fully veiled woman in northern Europe — vitamin D deficiency produces aching bones, weak proximal muscles (the thigh and shoulder ones), and the kind of fatigue that doctors usually attribute to depression or aging. That's rare. The version most readers carry around is invisible.
What it costs you, in order of certainty:
- Winter respiratory infections you'd otherwise dodge. Daily or weekly supplementation cuts the rate of colds, flu, and other respiratory infections — modestly in most adults, dramatically in the deeply deficient subgroup Martineau et al. 2017. If you're the person who gets two bad colds a winter and one of them turns into something that lingers for a month, this is the lever that quietly moves.
- Slow-burn autoimmune disease. Five years of supplementation cut new autoimmune diagnoses — rheumatoid arthritis, psoriasis, autoimmune thyroid, polymyalgia — by 22% in a 26,000-person trial Hahn et al. 2022. The version of you that gets a rheumatoid arthritis diagnosis at 55 isn't reading symptoms now.
- Years of life you'd otherwise have. The genetic-causation evidence from the UK Biobank is unusually clean: moving from a 25 nmol/L baseline to 50 nmol/L corresponded to a 20% drop in all-cause mortality risk Sutherland et al. 2022. This isn't a fitness-influencer claim; it's the cleanest causal-inference data the field has.
- Falls and fractures, if you're older. Past 65, deficiency raises fall risk; 700–1,000 IU/day cuts that risk by about a fifth and meaningfully reduces hip fracture in deficient cohorts Bischoff-Ferrari et al. 2009 Bischoff-Ferrari et al. 2005. A hip fracture at 75 still has a one-in-four mortality rate within a year.
None of this is dramatic week-to-week. The reader who's deficient and ignoring it doesn't notice a difference between Tuesday and Friday. The cost is in the version of their next decade where the winter coughs are slightly worse, the autoimmune diagnosis lands a year earlier, the recovery from a stumble takes weeks longer.
How to actually do this
The 2024 Endocrine Society guideline did something unusual: it retired the “test first” approach for most people. Routine blood testing is no longer recommended, and the old target of 30 ng/mL (75 nmol/L) is no longer endorsed as a treatment goal. The new advice is to supplement empirically if you're in a higher-risk group, and to skip both the test and the supplement if you're a healthy adult under 75 with a normal life Demay et al. 2024. The US Preventive Services Task Force reached a similar conclusion: there isn't enough evidence to recommend screening asymptomatic adults USPSTF 2021.
If you want a blood test — reasonable if you're in multiple risk groups or want a baseline — ask for 25-hydroxyvitamin D. A level above 50 nmol/L (20 ng/mL) is sufficient for almost every endpoint that matters. Below 30 nmol/L is the threshold for real deficiency.
The K2 question
The mechanistic argument for pairing D3 with vitamin K2 goes like this: D drives intestinal calcium absorption. Where that calcium ends up depends on a protein called matrix Gla-protein, which inhibits calcium deposition in the lining of arteries. Matrix Gla-protein only works in its activated form, and the activator is K2 (specifically the MK-7 form found in fermented foods like natto, aged cheeses, and the dark meat of some animals). Without enough K2, the theory runs, supplementing high-dose D might direct calcium toward arterial walls instead of bone.
The honest summary: the mechanism makes sense, the observational signal is striking, and the hard-endpoint randomised trial showing D3-plus-K2 beats D3-alone on cardiovascular mortality has not been done. The 2024 Endocrine Society guideline doesn't recommend it. At 1,000–2,000 IU/day of D3, the upside-vs-downside of adding 100–200 µg of MK-7 looks favourable: trivial cost, no documented risk in non-warfarin patients, plausible mechanism, supportive epidemiology. At 4,000+ IU/day of D3 the case is stronger, since the calcium-direction problem scales with the dose driving absorption.
If you're on warfarin or any other vitamin-K-antagonist anticoagulant: don't add K2 without your prescribing clinician. The entire mechanism of warfarin is K antagonism, and adding K disrupts the dose.
Sun as an alternative
Sun exposure does work for the people who live close to the equator and aren't darkly pigmented. The dermatology professional consensus is firmly against using sun as a deliberate vitamin D strategy: no UVB exposure dose has been identified that reliably builds vitamin D without raising skin cancer risk, and skin cancer is the most common cancer in the developed world. Sunscreens block the same wavelengths that make D Matsuoka et al. 1987, but applied at the amounts most people actually use, they don't fully shut down synthesis. Tanning beds are the worst of both worlds — you take on the melanoma risk without the rest of sunlight's effects.
The reasonable lay synthesis: get sun like a normal person, supplement orally for the rest.
What most guides get wrong
- “The big trials proved it doesn't work.” The big trials proved it doesn't work in already-replete adults. That's a very different statement from “doesn't work” — and the subgroups within those same trials who started deficient mostly did benefit. The autoimmune result from VITAL applied to the whole cohort, not just the deficient subgroup Hahn et al. 2022.
- “Once-a-month mega-dose is more convenient and just as good.” Not on the immune endpoint. The respiratory-infection effect appears with daily and weekly dosing and vanishes with monthly bolus regimens Martineau et al. 2017. The body wants a steady level, not a flood.
- “A good summer banks enough for the year.” The storage form has a half-life of around two months. A summer of robust outdoor exposure can carry someone past October; it cannot carry them through February at high latitude.
- “Higher is always better.” No. The mortality benefit in the genetic-causation data flattens above 50 nmol/L of 25-hydroxyvitamin D Sutherland et al. 2022. Above 4,000 IU/day taken chronically, the safety margin shrinks; sustained doses over 10,000 IU/day can cause hypercalcaemia Holick 2007. The dose-response curve plateaus and then turns harmful.
- “Sunscreen ruins vitamin D production.” In a lab, perfectly applied at SPF 15+, yes. In real-world use — thin coats, irregular coverage, hands and ears missed — cutaneous synthesis continues. Use sunscreen.
When not to do this
What changes if you correct it
For the genuinely deficient — the third of the population with serum levels below 50 nmol/L, concentrated in the high-latitude, dark-skinned, older, indoor-living, and pregnant — repleting is one of the lowest-friction, highest-leverage interventions in the catalogue.
- First month or two: serum levels rise quietly. If you were severely low, the aches and the foggy fatigue you might have been blaming on age or stress start lifting. If you weren't severely low, you notice nothing — and that's the expected outcome.
- First winter: the cold that would have knocked you out for two weeks is one that knocks you out for four days, or that you skip entirely Martineau et al. 2017. The effect is statistical, not guaranteed; the version of you that supplements catches fewer of them than the version that doesn't.
- One to two years: if you're older, fewer falls, fewer near-misses on the stairs Bischoff-Ferrari et al. 2009. Bone density in the deficient stops drifting downward.
- Five years: the autoimmune disease that would have landed on someone with your genetic profile lands on someone else with the same profile. Twenty-two percent reduction over five years Hahn et al. 2022 means in a cohort of 1,000 people destined for new rheumatoid arthritis, psoriasis, autoimmune thyroid or polymyalgia diagnoses, 220 of them get a different next decade.
- Decade and beyond: the all-cause mortality curve moves — not dramatically, but durably. The deficient version of you trades years; the replete version keeps them Sutherland et al. 2022.
If you're not deficient — outdoor work, lighter skin, lower latitude, normal weight, under 70 — you can stop reading. The same trials that show benefit for the deficient show essentially nothing for you. The cost of supplementing anyway is roughly $10 a year and zero side effects; the gain is roughly zero. The Endocrine Society's 2024 honest position is that the healthy 18-to-74 adult doesn't materially need this.
Related rabbit holes
- Vitamin K2 on its own — the cardiovascular case for menaquinone is interesting enough to warrant its own treatment, separate from the D-pairing question.
- Calcium supplementation — historically bundled with vitamin D, increasingly under question for its own cardiovascular signal. Worth a separate look before adding it on top.
- Sun exposure as a whole — the dermatology-vs-everything-else trade-off is bigger than vitamin D. Sunlight does things to mood, circadian rhythm, nitric oxide, and skin cancer that D alone doesn't capture.
- Magnesium status — vitamin D metabolism requires magnesium-dependent enzymes; chronic low magnesium can blunt the response to D supplementation.
- Omega-3 fatty acids — the other half of the VITAL trial. The autoimmune signal showed up for omega-3 too, though weaker.
- — A five-year trial showed daily vitamin D cut new autoimmune diagnoses by about a fifth — a cheap edge given how often these hit women.
- — Repleting deficient vitamin D cuts falls and fractures and is part of the bone-protection foundation.
- — Low vitamin D weakens bone; it's one of the levers behind a poor density scan.
- — Low vitamin D is tied to recurrent inner-ear vertigo (BPPV); topping it up measurably reduces repeat attacks of the spinning.
- — Untreated celiac blocks vitamin D absorption, so deficiency hangs on despite sun or pills until the gut actually heals.
- — Your body needs magnesium to activate vitamin D, so a magnesium gap can blunt what your D supplement actually does.
- — Sun on skin is how you'd normally make vitamin D, but above the 40th parallel the winter angle stops the chemistry.
- — K2 directs the calcium your D helps absorb into bone instead of artery wall; they're a natural pair.
- — Getting outside for your circadian dose also makes vitamin D, but far from the equator the winter sun is too weak, so a supplement covers the gap.
- — The same dark winters that drop your vitamin D drive seasonal depression — but light, not the pill, is the actual treatment.
- — Daily sunscreen filters the UVB your skin uses to make D. If you're already low at your latitude, get it from food or a pill, not burns.
Substance and claimed effects
The substance is vitamin D status — specifically, the seasonal and demographic deficiency driven by the geometry of UVB-driven cutaneous synthesis, and the supplementation choices that correct it. Above ~37°N (a line running through San Francisco, Richmond, southern Spain, central China) the sun's angle through the winter atmosphere strips out the 290–315 nm UVB band that converts 7-dehydrocholesterol in the skin to previtamin D3. The classic Webb-Kline-Holick photolysis experiment showed Boston (42.2°N) makes no cutaneous vitamin D from November through February; Edmonton (52°N) from October through March; sub-tropical latitudes (18–34°N) maintain synthesis year-round Webb et al. 1988. Melanin is a competing UVB absorber: darkly pigmented skin requires roughly 3–5× the exposure for the same precursor yield Holick 2007. The substance has two corrective routes — sun exposure (the evolutionary default; carries a skin-cancer trade-off) and oral cholecalciferol (with an open question about whether to pair it with K2). Claimed consequences across dimensions: bone density and fractures, vascular calcification, respiratory infection risk, autoimmune incidence, mood, mortality, type-2 diabetes progression, fall risk in older adults, and the skin-cancer trade-off in the synthesis pathway. The entry covers each.
Evidence by addressing question
Mechanism
Vitamin D is a prohormone, not a vitamin in the classical sense. UVB photons at 290–315 nm cleave the B-ring of 7-dehydrocholesterol in keratinocytes, yielding previtamin D3, which thermally isomerises to cholecalciferol over hours. Cholecalciferol is hydroxylated in the liver (CYP2R1) to 25-hydroxyvitamin D — the storage form measured in clinical assays — then in the kidney (CYP27B1) to 1,25-dihydroxyvitamin D, the active hormone that binds the vitamin D receptor (VDR). VDR is expressed in every nucleated cell; the receptor's tissue distribution explains why deficiency produces effects far beyond the calcium axis Holick 2007.
Latitude bottleneck: the further from the equator, the more atmosphere UVB photons must traverse at winter solar zenith. Above the ∼37th parallel, the path length attenuates UVB below the threshold needed for photolysis — the “vitamin D winter.” Pigmentation bottleneck: melanin is a broad-spectrum UV chromophore that competes with 7-dehydrocholesterol for the same photons. A constitutively dark-skinned individual at 50°N is doubly disadvantaged. The genetic architecture of skin colour (SLC24A5, SLC45A2, OCA2 variants) has been mapped onto vitamin D-deficiency risk in African American cohorts.
K2 hypothesis: vitamin D upregulates intestinal calcium absorption and osteoblast-derived matrix Gla-protein (MGP). MGP carboxylation — the activation step that lets MGP inhibit calcium deposition in soft tissue — requires vitamin K2 (menaquinone) as cofactor. Mechanistically, isolated high-dose D3 in a K2-insufficient adult could direct calcium toward arterial intima rather than bone. The mechanism is plausible; the clinical-trial confirmation is thin.
Evidence
The evidence base divides cleanly into observational/genetic data (consistently pointing at deficiency as harmful) and large randomised trials of supplementation (mostly null on hard endpoints, with three exceptions). The reconciliation is the deficiency threshold.
Observational and genetic. NHANES 2001–2018 found 33.6% of US adults below the 50 nmol/L (20 ng/mL) threshold and 71.7% below 75 nmol/L (30 ng/mL); a pooled analysis of 7.9 million participants worldwide found 15.7% below 30 nmol/L and 47.9% below 50 nmol/L Cui et al. 2023. The UK Biobank nonlinear Mendelian-randomisation analysis — the most rigorous causal-inference attempt to date — found an L-shaped mortality curve: odds of all-cause death increased by 25% (OR 1.25, 95% CI 1.16–1.35) at a genetically predicted 25(OH)D of 25 nmol/L versus 50 nmol/L; no further benefit above 50 nmol/L Sutherland et al. 2022.
RCT — bone, falls. The 2005 Bischoff-Ferrari meta-analysis pooled five fracture RCTs (n=9,294) and found 700–800 IU/day cholecalciferol reduced hip fracture by 26% and any non-vertebral fracture by 23%; 400 IU/day was insufficient Bischoff-Ferrari et al. 2005. Her 2009 fall-prevention meta-analysis (eight RCTs) found 700–1,000 IU/day reduced falls in adults ≥65 by 19% (RR 0.81), again with a 700 IU floor Bischoff-Ferrari et al. 2009. The 2022 VITAL ancillary on fractures was null in the general healthy adult population, but the cohort was largely vitamin D-replete at baseline LeBoff et al. 2022. Reconciliation: D corrects fractures when the cohort starts deficient; it does nothing when they don't.
RCT — cardiovascular, cancer. VITAL randomised 25,871 US adults (mean age 67, including 5,106 Black participants) to 2,000 IU/day vs placebo for a median 5.3 years. Null on total invasive cancer, null on major cardiovascular events (MI, stroke, CV death) Manson et al. 2019. ViDA, a New Zealand trial of monthly 100,000 IU bolus dosing in 5,110 adults, was also null on cardiovascular endpoints Scragg et al. 2017. DO-HEALTH, a European trial of 2,000 IU/day in 2,157 adults ≥70, was null on blood pressure, fracture, infection rate, cognition, and major cardiovascular events Bischoff-Ferrari et al. 2020. The trial cohorts were largely D-replete at baseline.
RCT — immunity. Martineau's 2017 individual-participant meta-analysis of 25 RCTs (n=10,933) showed daily/weekly vitamin D reduced acute respiratory tract infection (ARI) by 12% overall (OR 0.88); in baseline-deficient adults (<25 nmol/L) the reduction was 70% (OR 0.30, 95% CI 0.17–0.53). Bolus doses (≥30,000 IU monthly) did not work Martineau et al. 2017. The 2021 aggregate-data update with 46 trials (n=75,541) confirmed a modest protective effect (OR 0.92) Jolliffe et al. 2021.
RCT — autoimmunity. The VITAL autoimmune ancillary (Hahn 2022) is the strongest hard-endpoint signal from a primary-prevention vitamin D trial: 2,000 IU/day for 5.3 years reduced incident confirmed autoimmune disease (rheumatoid arthritis, polymyalgia rheumatica, autoimmune thyroid disease, psoriasis) by 22% (HR 0.78, 95% CI 0.61–0.99) Hahn et al. 2022.
RCT — mood, diabetes. VITAL-DEP (n=18,353) was null on incident depression and on continuous mood scores over 5.3 years Okereke et al. 2020. D2d randomised 2,423 prediabetic adults to 4,000 IU/day; null on progression to type 2 diabetes overall, with a per-protocol signal in adherent participants who achieved ≥125 nmol/L Pittas et al. 2019.
Protocol
Endocrine Society 2024 retired the previous “test, then treat to >30 ng/mL” framing — the guideline now recommends empirical supplementation without routine testing in defined groups, and explicitly does not endorse a target 25(OH)D threshold Demay et al. 2024. Recommended populations: adults ≥75 (mortality benefit), pregnant people (pre-eclampsia, neonatal mortality), high-risk prediabetics (D2d adherent subgroup), and children ≤18 (rickets and ARI prevention). Healthy adults 18–74 do not need to exceed the IOM RDA (600–800 IU) under the guideline; this is a departure from the 2011 Endocrine Society guideline.
USPSTF 2021 issued an I-statement (insufficient evidence) on screening asymptomatic adults USPSTF 2021. The practical implication: routine clinic-ordered 25(OH)D is not recommended; treat empirically with low-dose D3 in at-risk groups.
Dose-response: 1,000 IU/day raises serum 25(OH)D by roughly 25 nmol/L (10 ng/mL) at steady state, taking ~3 months. 2,000 IU/day is the VITAL dose and the de-facto standard for high-latitude adults wanting insurance. 4,000 IU/day is the IOM upper limit and the D2d dose; safe in trials but offers no documented additional benefit. D3 (cholecalciferol) is roughly 2–3× more potent than D2 (ergocalciferol) at raising and maintaining 25(OH)D — the D2 form has a markedly shorter serum half-life.
K2 co-supplementation: the Rotterdam Study (n=4,807, 10-year follow-up) found highest-tertile dietary menaquinone intake (MK-4 through MK-10) was associated with 41% lower coronary heart disease incidence and 52% lower severe aortic calcification versus lowest tertile, in an observational cohort Geleijnse et al. 2004. The 3-year RCT of MK-7 (180 µg/day) co-supplementation has shown reduced progression of arterial stiffness in postmenopausal women, but no large hard-endpoint trial has tested D3-with-vs-without-K2 on cardiovascular mortality. The mechanism is sound and the food intake correlation is striking; the prescription is “reasonable, especially at higher D3 doses” rather than guideline-endorsed.
Contraindications
Sarcoidosis and other granulomatous diseases dysregulate the 1-alpha-hydroxylase enzyme — macrophages produce active 1,25(OH)2D in an uncontrolled fashion, and supplementation precipitates hypercalcaemia. Primary hyperparathyroidism: D supplementation can worsen hypercalcaemia. Active hypercalcaemia of any cause. Sustained intake >10,000 IU/day risks vitamin D toxicity (hypercalcaemia, hypercalciuria, nephrocalcinosis) Holick 2007; below the 4,000 IU/day IOM UL, toxicity is essentially absent IOM 2011. Vitamin K2 co-supplementation interacts with warfarin (the entire mechanism of warfarin is K antagonism); patients on coumadin must coordinate with their clinician.
Misconceptions
“You can get enough D from sun in summer to last the year” — partially true. Adequate summer exposure can build serum 25(OH)D into the 75–125 nmol/L range; with a half-life of ~2 months, this carries a deficiency-naive person partway through winter, but doesn't bridge the full Dec–Feb gap above 40°N. “The window for D synthesis is during peak UV” — correct; UVA-only hours (morning, late afternoon, all of winter at high latitude) do not make D. “Sunscreen blocks vitamin D” — SPF≥15 in vitro reduces cutaneous D3 production by >95% Matsuoka et al. 1987; in practice, sunscreen is applied unevenly and at suboptimal amounts, so real-world serum 25(OH)D is largely unaffected. “If RCTs show null, supplementation is useless” — conflates baseline-replete RCT populations with the deficient minority who actually benefit; this is the central interpretive mistake.
Audience
Population variability is large enough that this entry warrants explicit audience scoping. Highest-risk groups: dark-skinned individuals at northern/southern latitudes (latitude–pigment mismatch); housebound or veiled adults; adults ≥75 (declining cutaneous synthesis and renal 1-alpha-hydroxylase activity); pregnant people; obese adults (sequestration of fat-soluble D in adipose tissue lowers bioavailable serum levels); patients with malabsorption (Crohn's, celiac, bariatric surgery); chronic kidney disease (impaired 1-alpha-hydroxylation). Lowest-risk: outdoor workers at low latitude with lightly pigmented skin.
Failure modes
The biggest practical failure mode is bolus dosing — the “50,000 IU once a month” prescription that doctors still write. Martineau's IPD showed bolus regimens do not produce the ARI-prevention signal that daily/weekly regimens do Martineau et al. 2017; the steady-state biology requires steady-state dosing. Second failure mode: supplementing only in winter. Steady-state 25(OH)D takes ~3 months to plateau; starting in November means peak coverage in February, by which time half the winter is over. Year-round daily 1,000–2,000 IU avoids the lag. Third: D2 instead of D3. D2 raises serum 25(OH)D less and for less time; the D3 form is preferred. Fourth: assuming a clinic serum 25(OH)D result is precise — inter-assay variability is ±25%, and free 25(OH)D (the biologically active fraction) is not routinely measured.
Practicalities
Generic cholecalciferol 1,000–2,000 IU softgels run roughly $5–15/year. K2 as MK-7 at 100–200 µg/day adds $30–60/year. Combined D3/K2 products are widely available. Lab cost (if a clinician orders 25(OH)D): $20–50 cash, often covered by insurance with indication. Free-25(OH)D assays exist (LabCorp, Quest) but are not standard.
Stakes
Deficiency is symptomatic at the extreme — osteomalacia, proximal muscle weakness, secondary hyperparathyroidism — and silent at the moderate end. The moderate-deficiency mortality signal from UK Biobank MR is the load-bearing data point for the “stakes” section: 25% higher all-cause mortality risk at genetically-predicted 25(OH)D of 25 vs 50 nmol/L Sutherland et al. 2022. ARI burden is the felt-experience anchor: 70% reduction in respiratory infections in the deficient strata is one of the largest effect sizes in any vitamin trial Martineau et al. 2017. Autoimmune disease prevention adds a slow-burn dimension — 22% over 5 years.
Payoff
For the genuinely deficient (~30% of US adults at <50 nmol/L, higher in dark-skinned and high-latitude subgroups), repletion produces: fewer winter respiratory infections within months Martineau et al. 2017; reduced fall risk in older adults at 700+ IU/day Bischoff-Ferrari et al. 2009; reduced fracture risk in older deficient adults Bischoff-Ferrari et al. 2005; 22% lower autoimmune disease incidence at 5-year horizon Hahn et al. 2022; mortality benefit at the population level Sutherland et al. 2022. For the already-replete, repayoff is null on every measured hard endpoint Manson et al. 2019 Bischoff-Ferrari et al. 2020.
Out of scope
Sun-exposure protocols as a deliberate D-loading strategy — the dermatology consensus (AAD) is that no safe UVB exposure threshold exists that boosts D without raising skin-cancer risk; oral D3 dominates the trade-off. Phototherapy lamps and tanning beds: industrial-strength UVB at narrow band raises D, but lifetime melanoma risk is real and disproportionate. K2 as a standalone cardiovascular intervention belongs in its own entry (the Rotterdam observational data deserves dedicated treatment).
The credibility range
Optimist case
Vitamin D is foundational, undermeasured, and underdosed across most of the developed world. Cutaneous synthesis fails for ~four months a year above 40°N; dark-skinned populations in those latitudes are functionally deficient most of the year. The Mendelian-randomisation evidence is the cleanest causal-inference design available outside RCTs, and the L-shaped mortality curve up to 50 nmol/L is unambiguous Sutherland et al. 2022. The VITAL autoimmune signal — 22% reduction over 5 years on a hard, adjudicated endpoint — is one of the most striking primary-prevention results from any nutritional RCT in decades Hahn et al. 2022. The Martineau ARI meta-analysis shows a 70% reduction in deeply deficient subgroups Martineau et al. 2017. Bone-and-fall benefits at 700–1,000 IU in older deficient adults are settled Bischoff-Ferrari et al. 2005 Bischoff-Ferrari et al. 2009. The reason large general-population RCTs are null is that they recruit cohorts already replete; the deficient subgroups within them do show benefit (LeBoff bone-density signal in baseline-low subset; D2d adherent subgroup; Martineau strata). 1,000–2,000 IU/day of cholecalciferol is essentially free, essentially zero-risk below 4,000 IU, and corrects an exposure mismatch most of the population has. Adding K2 (MK-7 100–200 µg) covers the mechanistic concern about misdirected calcium with similarly trivial cost.
Skeptic case
The largest RCTs ever conducted on hard endpoints — VITAL, ViDA, DO-HEALTH, D2d, VITAL-DEP — are all null on their primary outcomes Manson et al. 2019 Scragg et al. 2017 Bischoff-Ferrari et al. 2020 Pittas et al. 2019 Okereke et al. 2020. The LeBoff fracture ancillary in the general population is null LeBoff et al. 2022. The 2024 Endocrine Society guideline explicitly walked back the “treat to 30 ng/mL” framing for healthy adults under 75 and stopped endorsing any target threshold Demay et al. 2024; USPSTF gives an I-statement on screening USPSTF 2021. Observational deficiency-mortality associations are vulnerable to reverse causation — sick people stay indoors, are less mobile, eat worse, weigh more — and even MR analyses depend on model assumptions about gene-environment interaction. Subgroup signals in trials are post-hoc and underpowered. The autoimmune VITAL result needs replication; one trial does not move guideline practice. The K2 story is observational (Rotterdam) plus mechanism; no hard-endpoint RCT exists. The supplementation industry has incentives to amplify the optimist case.
Author's call
This entry lands skewed-optimist with explicit conditioning on deficiency status. The Mendelian-randomisation mortality signal, the autoimmune RCT result, the deficient-strata ARI effect, and the fall/fracture data in older adults at 700–1,000 IU are real and convergent. The null hard-endpoint trials are not the rebuttal they look like — they recruited replete cohorts and were under-powered to detect deficient-subgroup benefit. The Endocrine Society 2024 position — supplement empirically without testing in defined risk groups, accept that healthy 18–74 likely don't benefit — is the most defensible synthesis. The headline action for the catalogue's reader: identify whether you sit in a risk group (latitude, pigmentation, age, body-weight, sun-exposure pattern), and supplement 1,000–2,000 IU/day of D3 with optional K2 if so. Controversy stays elevated (3/5) because the field is mid-realignment between deficiency-focused and general-population frames.
Stakeholder and incentive map
Supplement industry: incentive to amplify deficiency prevalence and broad-population recommendations; the D market is large and undifferentiated, with K2 as the upsell. Endocrine professional societies: Endocrine Society 2024 walked back its own 2011 thresholds, suggesting limited capture. USPSTF: structurally conservative, weights RCT-only evidence high. Public-health establishment (IOM, WHO): set the RDA on bone outcomes alone, conservative on upper limits. Dermatology (AAD): consistently opposes sun-as-D-source on cancer grounds, irrespective of the D trade-off. Functional medicine/wellness practitioners: aggressive D testing and supplementation, often well above guideline doses. Pharmaceutical industry: minimal stake — D3 is generic, cheap, and not promotable. Researchers who founded the field (Holick, Bischoff-Ferrari) have advocacy positions; their meta-analyses tilt optimist.
Population variability
Latitude: above ~40°N or below ~40°S, October–March effectively produces no cutaneous D. Below ~30°N/S, year-round synthesis is feasible. Skin pigmentation: Fitzpatrick I–II skin requires ~15 min midday summer exposure of arms+face for a daily dose; Fitzpatrick V–VI requires 60–90 min — impractical for most modern lifestyles. African Americans show the highest US VDD prevalence (>70% at <50 nmol/L in some NHANES strata). Age: cutaneous 7-dehydrocholesterol decreases with age; renal 1-alpha-hydroxylase activity declines; older adults make 25–50% less D from equivalent UVB exposure. Obesity: BMI ≥30 sequesters D in adipose; serum 25(OH)D is correspondingly lower for the same intake. Pregnancy: requirements rise; deficiency linked to pre-eclampsia and neonatal mortality (Endocrine Society 2024 recommends supplementation). Malabsorptive states (Crohn's, celiac, bariatric surgery, cystic fibrosis): higher doses required. Chronic kidney disease: impaired conversion to 1,25(OH)2D; calcitriol or paricalcitol is sometimes substituted under nephrology supervision.
Knowledge gaps
The pivotal trial that doesn't exist: a baseline-deficient cohort randomised to D3 repletion vs placebo, with hard endpoints. VITAL and DO-HEALTH excluded the deficient by design (ethics — can't randomise to known deficiency). The autoimmune result needs independent replication. K2 co-supplementation lacks any hard-endpoint RCT; the Rotterdam dietary signal could be confounded by everything correlated with fermented-dairy / liver / cheese consumption. The optimal 25(OH)D target is unknown — UK Biobank MR data points to 50 nmol/L as the threshold beyond which benefit plateaus, but the curve's resolution is coarse. The mechanism behind autoimmune prevention (T-reg modulation? cathelicidin? something else?) is hypothesised but not confirmed. Free 25(OH)D may be a better biomarker than total 25(OH)D, especially in obesity and pregnancy where binding-protein levels shift, but assays are not standardised.
The brief named latitude, pigmentation, supplementation (with/without K2), bone density, vascular calcification, immunity, mood, skin cancer risk, and RCT hard endpoints. The article covers all of those, though several with narrow treatment: vascular calcification is covered through the K2 section rather than a dedicated piece, and the skin-cancer-vs-synthesis trade-off is one paragraph in the alternatives section. Editorial judgement: a long disquisition on the sun-vs-skin-cancer trade-off was tempting but would have hijacked the article's centre of gravity, which is the deficiency-treatment decision. Worth surfacing separately as a sun-exposure entry; flagged.
- Scoring difficulty: mood. Scored 1 despite the null VITAL-DEP primary outcome Okereke 2020 because the deficient-subgroup mechanism is real and the observational literature is consistent. A 0 would have under-rated. The risk of scoring 1 is over-rating; on balance, 1 is the honest number.
- Scoring difficulty: longevity. The UK Biobank MR data Sutherland 2022 is strong, but it's MR, not RCT. The autoimmune VITAL result Hahn 2022 nudges the score up. Settled on 3 (meaningful) rather than 4 (major) because the benefit is conditional on baseline deficiency.
- Controversy at 3. The 2024 Endocrine Society guideline materially retreated from its 2011 position. Endocrinology, USPSTF, and the dermatology community are not aligned on testing, supplementation thresholds, or sun guidance. 3 (active debate among reasonable experts) over 2 (minor pushback).
- K2 framing. No hard-endpoint RCT exists for D3-plus-K2 vs D3-alone on cardiovascular outcomes. The Rotterdam observational data and the MGP mechanism are genuine but not guideline-endorsed. The article positions K2 as “reasonable, especially at higher D3 doses” rather than “recommended.” The warfarin warning is the practical hazard.
- Skin cancer. Covered through the AAD position in the alternatives section. The catalogue likely warrants a dedicated sun exposure entry that addresses melanoma risk, nitric oxide effects, circadian alignment, and the full UVA/UVB trade-off — this entry shouldn't carry that weight.
- Beauty (direct) at 0. No documented short-term skin/face/hair effect from D supplementation per se. Acne and psoriasis links exist but are downstream of the immunity / autoimmune axis already scored. Beauty (cumulative) at 1 captures the bone-density-over-decades angle modestly.
- Future links to add when the entries exist: vitamin K2, sun exposure, calcium supplementation, magnesium status, omega-3 fatty acids.
- Contraindication tokens. Used
kidney-diseasefor the CKD case (D activation is renal) andblood-thinnerswould have applied to the K2 warfarin warning, but the contraindication is on the optional K2, not on D itself. Sarcoidosis and primary hyperparathyroidism are real contraindications but no closed-vocabulary token covers them; called out in the warning callout instead. - What was left out deliberately: COVID-19 outcomes (the literature exploded then partially retracted itself; not load-bearing for the catalogue); calcium co-supplementation (warrants its own entry given its own cardiovascular signal); detailed assay-variability discussion (too in-the-weeds for the reader); the older Endocrine Society 2011 framework (superseded).
Vitamin D Deficiency at Latitude
Generic D3 1,000-2,000 IU softgels run $5-15/year; adding K2 MK-7 costs an additional $30-60/year. Trivial.
One pill a day. Setup is trivial; the only ongoing effort is remembering to take it.
Multiple large RCTs (VITAL n=25,871, ViDA n=5,110, DO-HEALTH n=2,157, D2d n=2,423), individual-participant meta-analyses (Martineau 2017 IPD, Bischoff-Ferrari 2005/2009), Mendelian-randomisation causal-inference data (Sutherland 2022 UK Biobank), and recent professional guideline (Endocrine Society 2024). The body of evidence is large; the interpretation is contested. Solid 4, not 5 because the deficient-subgroup vs general-population reconciliation is not settled.
UK Biobank nonlinear MR shows 25% higher all-cause mortality at genetically predicted 25(OH)D of 25 vs 50 nmol/L (Sutherland 2022). VITAL autoimmune-disease HR 0.78 over 5y (Hahn 2022). Endocrine Society 2024 endorses supplementation in adults >=75 for mortality. Meaningful, replicated, but conditional on deficiency.
In baseline-deficient adults, repletion can reduce musculoskeletal pain and fatigue within weeks (Holick 2007). In already-replete adults VITAL and DO-HEALTH found no felt-experience changes. Holistic call: 2 (modest, deficiency-conditional).
Severe deficiency presents with proximal muscle weakness and fatigue; repletion resolves it. Mild-to-moderate deficiency shows no consistent energy signal in RCTs. Real but small contribution at the population level (Holick 2007).