Vitamin D: Does It Actually Work?
Last reviewed: 2026-02-21 00:00:00 +0000 UTC
Meta-analysis verifiedSnakeOilCheck earns commissions from qualifying purchases made through affiliate links on this site. Our editorial content and meta-analyses are produced independently and are not influenced by affiliate relationships.
What Is Vitamin D?
Vitamin D is a fat-soluble vitamin your body makes when sunlight hits your skin. It’s also found in fatty fish, egg yolks, and fortified foods. But most people don’t get enough from food and sun alone.
About 42% of American adults are deficient. That number jumps higher if you have dark skin, live far from the equator, or spend most of your time indoors.
You’ll see vitamin D marketed for everything from bone health to cancer prevention to muscle growth. Some of those claims hold up. Most depend heavily on whether you’re deficient or not.
We analyzed 4 major meta-analyses and systematic reviews covering 162 individual studies. Here’s what the data actually shows.
The Evidence, Claim by Claim
The sections below break down each claim with real numbers from published meta-analyses. Every effect size, confidence interval, and study count comes from peer-reviewed research.
Bones and Fractures: Not What You’d Expect
This is the claim everyone assumes is settled. It isn’t.
Bolland et al. reviewed 81 randomized controlled trials with over 53,000 people. Vitamin D supplementation alone didn’t significantly reduce fractures (RR = 0.97). It didn’t reduce falls either.
There’s a catch. When you combine vitamin D with calcium, elderly people do see a small reduction in hip fractures. And high-dose bolus vitamin D (one mega-dose per year) actually increased falls and fractures in one major trial.
The takeaway: vitamin D alone won’t protect your bones if your levels are already normal. If you’re deficient and elderly, it might help, especially with calcium.
Immune Function: The Strongest Claim
This is where vitamin D shines. Jolliffe and Martineau ran an individual participant data meta-analysis. That’s the gold standard. They pooled data from 43 trials and nearly 49,000 people.
The results: vitamin D reduced acute respiratory infections by about 8% overall (OR = 0.92). But the real story is in the subgroups.
People with very low vitamin D (under 25 nmol/L) saw a 37% reduction in infections. Daily dosing worked. Monthly or yearly bolus dosing didn’t work at all.
If you’re deficient, daily vitamin D is one of the most evidence-backed ways to support your immune system.
Depression: Small But Real
Cheng et al. pooled 25 studies with about 7,500 people. They found vitamin D modestly reduced depression scores (SMD = -0.28). That’s a small effect.
It got bigger in two groups: people with clinical depression and people who were vitamin D deficient. The prediction interval crosses zero, meaning not every study agrees.
Don’t count on vitamin D to fix depression. But if you’re deficient and struggling with low mood, correcting it might help as part of a larger treatment plan.
Muscle Strength: Probably Not
Tomlinson et al. looked at 13 studies with about 2,200 people. The pooled effect wasn’t significant (SMD = 0.17, p = 0.09). The confidence interval crossed zero.
The only positive signals came from severely deficient elderly people. If your vitamin D levels are normal, supplementing won’t make you stronger.
This claim doesn’t hold up for the general population.
D3 vs D2: Which Form Is Better?
Go with D3 (cholecalciferol). Research shows D3 raises blood levels about 70% more than D2 (ergocalciferol).
D3 is what your skin makes from sunlight. It’s the natural human form. D2 comes from plants and irradiated fungi.
Most supplements now use D3. Some vegan options use D3 sourced from lichen. D2 is fine if it’s your only option, but you’ll need a higher dose to get the same blood level increase.
How Much Do You Need?
The answer depends on your starting level. Get a 25(OH)D blood test first.
- Deficient (under 20 ng/mL): Your doctor may recommend 4000-5000 IU daily for 8-12 weeks.
- Insufficient (20-30 ng/mL): 2000-4000 IU daily to get into the optimal range.
- Sufficient (30-50 ng/mL): 1000-2000 IU daily to maintain levels.
- Above 50 ng/mL: You probably don’t need to supplement.
Take it with a meal that contains fat. Vitamin D is fat-soluble. Taking it on an empty stomach reduces absorption.
Obese individuals often need 2-3 times the standard dose because vitamin D gets stored in fat tissue.
Who Should NOT Take Vitamin D
Don’t take vitamin D if you have hypercalcemia (high blood calcium). Vitamin D increases calcium absorption, which makes this worse.
Avoid it if you have severe kidney disease. Your kidneys can’t convert vitamin D to its active form properly. Your doctor will prescribe the active form (calcitriol) instead if needed.
Be cautious with sarcoidosis and other granulomatous diseases. These conditions make your body extra sensitive to vitamin D, raising the risk of high calcium levels.
If you take digoxin, talk to your doctor before supplementing. Vitamin D-induced calcium changes can make digoxin toxic.
The Bottom Line
Vitamin D is one of the most studied supplements in the world. The evidence is clear but nuanced.
For immune support, it works, especially if you’re deficient. Daily dosing of D3 at 1000-4000 IU reduces respiratory infections. The lower your starting levels, the bigger the benefit.
For bones, it’s not the miracle the marketing suggests. Vitamin D alone doesn’t prevent fractures in most people. Combined with calcium, it might help the elderly.
For depression and muscle strength, the evidence is weaker. Small benefits exist for deficient people, but don’t expect much if your levels are already normal.
The real question isn’t “should I take vitamin D?” It’s “am I deficient?” Get tested. If you are, supplement with D3 daily. If you aren’t, you probably don’t need it.
The Evidence, Claim by Claim
Strengthens bones and prevents fractures ? Maybe
81 studies with over 53,000 people found vitamin D doesn't significantly reduce fracture risk on its own. The risk ratio of 0.97 means almost no difference from placebo. Combined with calcium, there's a small benefit for hip fractures in elderly people. Bolus (mega-dose) vitamin D may actually increase falls and fractures.
This is a very large evidence base. Low heterogeneity (I2 = 12.3%) means studies agree with each other. They agree that vitamin D alone doesn't prevent fractures in most people. Deficient individuals may still benefit.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.42 | no significant asymmetry detected |
|---|---|---|
| Trim & Fill | 2 imputed studies | Adjusted estimate: g = 0.98 |
| Fail-safe N (Rosenthal) | 120 studies needed to nullify result | |
Boosts immune function ✓ Works
43 studies with nearly 49,000 people found vitamin D reduces the odds of catching acute respiratory infections by about 8%. That's a real but modest effect. The big finding: people who start out very deficient (under 25 nmol/L) get a 37% reduction. Daily dosing works. Bolus dosing doesn't.
This is an individual participant data meta-analysis, the gold standard. Moderate heterogeneity (I2 = 38.2%) is acceptable. The key moderator finding is clear: baseline status matters. If you aren't deficient, the benefit is tiny.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.31 | no significant asymmetry detected |
|---|---|---|
| Trim & Fill | 1 imputed studies | Adjusted estimate: g = 0.93 |
| Fail-safe N (Rosenthal) | 680 studies needed to nullify result | |
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| <25 nmol/L | 8 | 0.63 |
| 25-50 nmol/L | 14 | 0.89 |
| >50 nmol/L | 21 | 0.97 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Daily | 28 | 0.83 |
| Weekly | 8 | 0.91 |
| Bolus | 7 | 1.02 |
Reduces depression ? Maybe
25 studies with about 7,500 people found a small reduction in depression scores with vitamin D. The effect (SMD = -0.28) is small but statistically significant. Bigger effects show up in people with clinical depression and in those who are vitamin D deficient. The prediction interval crosses zero, so not every new study will find a benefit.
High heterogeneity (I2 = 71%) means results vary a lot between studies. Some show clear benefits, others show nothing. Borderline publication bias is a concern. The effect may shrink as more studies come in.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.08 | borderline asymmetry suggesting possible bias |
|---|---|---|
| Trim & Fill | 3 imputed studies | Adjusted estimate: g = -0.18 |
| Fail-safe N (Rosenthal) | 95 studies needed to nullify result | |
Improves muscle strength ✗ No Evidence
13 studies with about 2,200 people found no significant effect of vitamin D on muscle strength. The effect size is tiny (SMD = 0.17) and the confidence interval crosses zero. The only positive results came from studies of severely deficient elderly people. If your vitamin D levels are normal, don't expect strength gains.
Moderate heterogeneity (I2 = 55%) reflects the split: deficient people sometimes benefit, replete people don't. Overall, the pooled effect isn't significant. This claim doesn't hold up for the general population.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.22 | no significant asymmetry detected |
|---|
Reduces respiratory infections ✓ Works
43 trials with nearly 49,000 people found vitamin D reduces respiratory infections by about 8% overall. Daily dosing at 400-1000 IU works best. Bolus (mega-dose) vitamin D doesn't work at all. The biggest benefit goes to people who are very deficient, where daily dosing cuts infection risk by 70%. An earlier gold-standard IPD meta-analysis of 25 trials confirmed these findings.
Based on two large meta-analyses (Jolliffe 2021, Martineau 2017 IPD). Moderate heterogeneity (I2 = 35.6%) is acceptable. A 2025 critique raised concerns about small-trial bias inflating the effect. After statistical correction, the overall effect becomes borderline. The deficiency subgroup remains robust.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | some funnel asymmetry noted, possibly small-trial bias |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Daily/weekly | 19 | 0.78 |
| Bolus | 14 | 1.01 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| 400-1000 IU | 10 | 0.7 |
| 1001-2000 IU | 0.9 | |
| >2000 IU | 0.98 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| <25 nmol/L | 6 | 0.3 |
| >=25 nmol/L | 6 | 0.75 |
Reduces cancer mortality ? Maybe
14 trials with over 104,000 people found vitamin D reduces cancer death risk by 6%, but that wasn't quite statistically significant. The key finding: daily dosing reduced cancer mortality by 12% (statistically significant). Bolus dosing did nothing. This is about cancer survival, not cancer prevention. Vitamin D doesn't stop cancer from developing, but daily supplementation appears to help people survive it.
Based on the largest IPD meta-analysis to date (Kuznia 2023). Zero heterogeneity (I2 = 0%) means all trials agree. The overall effect just misses significance. The daily dosing subgroup is significant (p = 0.019) but is technically a subgroup analysis. An earlier BMJ meta-analysis (Zhang 2019) found a significant 16% reduction before the negative D-Health bolus trial was included.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.6 | no significant asymmetry detected |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Daily | 10 | 0.88 |
| Bolus | 4 | 1.07 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| US/Europe | 9 | 0.87 |
| Other | 5 | 1.12 |
Prevents type 2 diabetes in prediabetes ? Maybe
11 trials with about 5,200 people who had prediabetes found vitamin D reduces the risk of developing type 2 diabetes by about 10%. That's a modest but statistically significant effect. The benefit was stronger in non-obese people (27% reduction) than in obese people (no significant effect). Vitamin D also increased the chance of reverting from prediabetes back to normal blood sugar by 24-48%.
Consistent finding across three separate meta-analyses (Zhang 2020, Sim 2024, Tobias 2025). Zero heterogeneity (I2 = 0%) is reassuring. The largest single trial (D2d, n=2,423) just missed significance on its own. This is specific to prediabetes. There's no evidence vitamin D prevents diabetes in the general population.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | no significant heterogeneity or publication bias observed |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Non-obese (BMI <30) | 0.73 | |
| Obese (BMI >=30) | 0.95 |
Improves blood sugar control in type 2 diabetes ? Maybe
39 trials with about 3,000 people with type 2 diabetes found vitamin D significantly reduces HbA1c by 0.30%, fasting blood sugar by 0.49 mmol/L, and insulin resistance (HOMA-IR by 0.39). These are modest but clinically meaningful improvements. The effects are strongest in people who are vitamin D deficient, overweight, or have poorly controlled diabetes (HbA1c above 8%).
Based on Chen 2024, the most comprehensive meta-analysis of glycemic outcomes in T2D. However, a 2026 meta-analysis (Gong, 10 RCTs) found no significant HbA1c effect using different methods and fewer studies. The discrepancy likely reflects differences in study selection and effect measure (WMD vs SMD). The weight of evidence favors a real but small benefit, mainly in deficient individuals.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | not explicitly reported |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Deficient | ||
| Sufficient |
Reduces risk of falls in older adults ✓ Works
35 trials with nearly 59,000 older adults found vitamin D at 800-1000 IU daily reduces fall risk by about 15%. The effect is strongest in people who are deficient, where fall risk drops by 31%. But dose matters a lot. Higher doses (above 1000 IU) don't help and may actually increase fall risk. Annual mega-doses are harmful. One large trial found a 500,000 IU annual dose increased falls by 15%.
Low heterogeneity (I2 = 11%) means studies agree. The dose-response is U-shaped: 800-1000 IU daily is the sweet spot. Large negative trials (DO-HEALTH, ViDA) enrolled mostly vitamin D-sufficient people, which explains their null results. The verdict 'works' applies specifically to 800-1000 IU daily in deficient older adults.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | not formally reported in network meta-analysis, low heterogeneity suggests consistent findings |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| 800-1000 IU/d | 0.78 | |
| >1000 IU/d | 1.05 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| <=50 nmol/L | 0.69 | |
| >50 nmol/L | 1 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Daily 800-1000 IU | 0.78 | |
| Annual bolus | 1.15 |
Reduces inflammation (CRP and cytokines) ? Maybe
An umbrella meta-analysis pooling 23 existing meta-analyses found vitamin D significantly reduces CRP (a key inflammation marker) and TNF-alpha. But IL-6 reduction wasn't significant. The biggest catch: the effect depends heavily on who's taking it. People with diabetes see a meaningful CRP drop of 0.67 mg/L. Healthy overweight people see zero effect. The VITAL trial (n=1,054) found only a temporary CRP reduction at year 2 that vanished by year 4.
The umbrella meta-analysis (Moslemi 2022) pools diverse populations, which inflates the overall effect. When you look at individual populations, the picture fragments. Disease populations (diabetes, dialysis) benefit. Healthy people don't. This is a 'maybe' because the effect is real but only in specific groups.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | not uniformly assessed across constituent meta-analyses |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| Abnormal glucose (T2DM, GDM) | 38 | -0.67 |
| Healthy/obese without comorbidities | 11 | 0.01 |
| Cancer patients | 8 | -0.09 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| CRP | -0.42 | |
| TNF-alpha | -0.27 | |
| IL-6 | -0.35 |
Improves sleep quality ? Maybe
A meta-analysis of 3 small trials with about 200 people found vitamin D improved sleep quality scores by 2.33 points on the PSQI scale. That's clinically meaningful (anything above 1.5 points matters). But there's a big problem: all 3 trials came from Iran, used high biweekly doses, and enrolled people with existing sleep or health issues. The largest sleep trial (VITAL-DEP, n=18,353) found no benefit. One well-designed US trial in postmenopausal women found vitamin D actually worsened sleep.
Zero heterogeneity (I2 = 0%) sounds impressive but reflects that all 3 pooled studies are remarkably similar (same country, similar doses, similar populations). Four other meta-analyses (2019-2022) also found significant PSQI improvements but with high heterogeneity (I2 = 75-85%). Short-term benefits (8-12 weeks) don't appear in longer trials. GRADE: moderate certainty.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = — | not assessed, only 3 studies in pooled analysis |
|---|
Lowers blood pressure ✗ No Evidence
46 trials with over 4,500 people found vitamin D has zero effect on blood pressure. The gold-standard IPD meta-analysis found 0.0 mmHg change in systolic BP and -0.1 mmHg in diastolic BP. An umbrella meta-analysis of 21 meta-analyses found a barely significant 0.69 mmHg reduction, which is clinically meaningless. Observational studies show a strong link between low vitamin D and high blood pressure, but RCTs don't support a causal relationship.
This is one of the most definitive null findings in the vitamin D literature. Low heterogeneity (I2 = 21%) means studies consistently agree: no effect. A Mendelian randomization study (natural genetic experiment) also found no causal link. The only subgroup with a real signal is hypertensive people who are severely vitamin D deficient, where SBP drops about 6.6 mmHg. But this is correcting a deficiency, not a pharmacological effect.
View full statistical analysis
Publication Bias Assessment
| Egger's Test | z = —, p = 0.6 | no significant asymmetry in Zhang 2020 SBP analysis |
|---|
Subgroup Analysis
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| General population | 27 | 0 |
| Hypertensive + deficient | -6.58 |
| Subgroup | Studies (k) | Effect (g) |
|---|---|---|
| IPD (27 trials) | 27 | -0.5 |
| Aggregate (46 trials) | 46 | 0 |
Dosage Guide
| Dose Range in Studies | 1000-4000 IU daily |
|---|---|
| Most-Studied Dose | 2000 IU daily for most adults |
| Best Form | D3 (cholecalciferol) preferred over D2 (ergocalciferol) |
| Timing | With a meal containing fat for better absorption |
| Time to Effect | 4-8 weeks to significantly change blood levels |
| Cycling | No cycling needed. Year-round use, especially in winter. |
| Notes | Get your 25(OH)D level tested. Target 30-50 ng/mL (75-125 nmol/L). D3 raises blood levels about 70% more than D2. Obese individuals may need 2-3x the standard dose. |
Ask Your Doctor Before Taking If You Have
- Hypercalcemia (high blood calcium)
- Severe kidney disease (impaired vitamin D metabolism)
- Sarcoidosis and other granulomatous diseases (increased sensitivity)
- Williams syndrome (increased calcium sensitivity)
Drug Interactions
| Medication | Risk | Why |
|---|---|---|
| Thiazide diuretics | moderate | Both raise calcium levels, increasing hypercalcemia risk |
| Digoxin | high | Vitamin D-induced hypercalcemia can cause dangerous digoxin toxicity |
| Corticosteroids | moderate | Steroids reduce calcium absorption and may lower vitamin D levels |
| Orlistat and cholestyramine | moderate | Reduce fat absorption, which lowers vitamin D uptake |
Possible Side Effects
- Hypercalcemia at very high doses (above 10,000 IU daily long-term)
- Nausea and vomiting at toxic levels
- Kidney stones (slight risk increase, especially with calcium)
Products That Match the Research
We're still verifying product links for this supplement. Check back soon.
What to Avoid
D2 is less effective than D3 at raising blood levels, needs higher doses
Frequently Asked Questions
Should I take vitamin D3 or D2?
D3 (cholecalciferol) is the better choice. Research shows D3 raises blood levels about 70% more effectively than D2 (ergocalciferol). D3 is the form your skin makes from sunlight. D2 comes from plants and fungi. Unless you're vegan and can't find vegan D3, go with D3.
How much vitamin D should I take daily?
Most adults do well with 1000-2000 IU daily. If you're deficient (under 20 ng/mL), your doctor might recommend 4000-5000 IU daily for 8-12 weeks to bring levels up. The safe upper limit is 4000 IU daily for long-term use. Get your blood level tested to know your starting point.
Does vitamin D prevent colds and flu?
It helps, but the size of the benefit depends on your starting levels. A large meta-analysis of 43 trials found an 8% reduction in respiratory infections overall. People who were very deficient saw a 37% reduction. Daily dosing works better than taking a big dose once a month.
Can vitamin D improve my mood?
Maybe, especially if you're deficient. A meta-analysis of 25 studies found a small but significant reduction in depression scores. The effect was stronger in people with clinical depression and low vitamin D levels. It's not a replacement for therapy or antidepressants, but correcting a deficiency might help.
Can vitamin D reduce cancer risk?
Not exactly. The pooled evidence from 14 large trials with over 104,000 people suggests vitamin D doesn't prevent cancer. But it may improve survival if you do get cancer. Daily dosing reduced cancer death risk by 12% in a major meta-analysis. Bolus dosing didn't help. This isn't a reason to take vitamin D for cancer prevention, but it's worth knowing if you're already supplementing.
Does vitamin D help prevent type 2 diabetes?
If you have prediabetes, possibly. The pooled evidence from 11 trials with about 5,200 prediabetic people found a 10% reduction in the risk of developing type 2 diabetes. Non-obese people saw a bigger benefit (27% reduction). Vitamin D also increased the chance of reverting from prediabetes back to normal blood sugar. But this doesn't apply to the general population.
Can vitamin D lower blood sugar in diabetes?
Based on the available evidence from 39 trials, vitamin D appears to modestly reduce HbA1c by about 0.30% and fasting blood sugar by about 8.8 mg/dL in people with type 2 diabetes. The benefit is strongest in people who are vitamin D deficient or have poorly controlled diabetes. It's a complement to standard treatment, not a replacement.
Does vitamin D prevent falls in older adults?
Based on the available evidence from 35 trials with nearly 59,000 people, vitamin D at 800-1000 IU daily appears to reduce fall risk by about 15% in older adults. The benefit is strongest in people who are vitamin D deficient (31% reduction). But dose matters. Higher doses don't help and may actually increase fall risk. Annual mega-doses are harmful.
Does vitamin D reduce inflammation?
It depends on who you are. An umbrella meta-analysis of 23 meta-analyses found vitamin D reduces CRP (a key inflammation marker) significantly. But the effect is strongest in people with diabetes or other inflammatory conditions. In healthy people, large trials like VITAL found little to no effect. If you're otherwise healthy with normal vitamin D levels, don't expect much anti-inflammatory benefit.
Can vitamin D improve sleep?
Maybe. A meta-analysis found a 2.33-point improvement on the PSQI sleep scale, which is clinically meaningful. But only 3 small trials were pooled, all from Iran with high-dose biweekly dosing. The largest sleep trial (over 18,000 people) found no benefit. If you're deficient and have sleep problems, correcting your vitamin D might help. But it's not a reliable sleep aid.
Does vitamin D lower blood pressure?
No. The pooled evidence from 46 trials with over 4,500 people found zero effect on blood pressure (0.0 mmHg change). Observational studies show a link between low vitamin D and high blood pressure, but randomized trials don't support supplementing to lower BP. The only exception: people who are both hypertensive and severely vitamin D deficient may see a small drop when correcting the deficiency.
Can you take too much vitamin D?
Yes. Vitamin D toxicity is rare but real. It usually happens above 10,000 IU daily for months. Symptoms include nausea, weakness, and dangerously high calcium levels. The safe upper limit is 4000 IU daily for most adults. Always test your blood levels if you're taking high doses.
Want to see the data? We summarize the published research and show you the pooled data from randomized controlled trials. Read our full methodology and dataset below
Summary
Based on our systematic summary of 12 health claims across 376 studies with 336,860 total participants, 3 claims have strong evidence supporting them, 7 claims show promising but incomplete evidence, and 2 claims lack sufficient evidence. Evidence certainty ranges from Grade A (strong) to Grade D (insufficient) across claims.
Summary of Findings
| Outcome | Studies | Participants | Effect Size (95% CI) | Certainty |
|---|---|---|---|---|
| Strengthens bones and prevents fractures | 81 | 53,537 | RR 0.97 (0.93 to 1.02) | Grade B |
| Boosts immune function | 43 | 48,488 | OR 0.92 (0.86 to 0.99) | Grade A |
| Reduces depression | 25 | 7,534 | SMD -0.28 (-0.53 to -0.03) | Grade B |
| Improves muscle strength | 13 | 2,205 | SMD 0.17 (-0.03 to 0.37) | Grade C |
| Reduces respiratory infections | 43 | 48,488 | OR 0.92 (0.86 to 0.99) | Grade A |
| Reduces cancer mortality | 14 | 104,727 | RR 0.94 (0.86 to 1.02) | Grade B |
| Prevents type 2 diabetes in prediabetes | 11 | 5,221 | RR 0.9 (0.81 to 0.99) | Grade B |
| Improves blood sugar control in type 2 diabetes | 39 | 2,982 | WMD -0.3 (-0.43 to -0.18) | Grade B |
| Reduces risk of falls in older adults | 35 | 58,937 | RR 0.85 (0.74 to 0.95) | Grade A |
| Reduces inflammation (CRP and cytokines) | 23 | — | ES -0.42 (-0.55 to -0.29) | Grade B |
| Improves sleep quality | 3 | 200 | WMD -2.33 (-3.09 to -1.57) | Grade B |
| Lowers blood pressure | 46 | 4,541 | WMD 0 (-0.8 to 0.8) | Grade D |
Review Protocol
For each claim, we searched for the most recent published systematic review or meta-analysis of randomized controlled trials evaluating vitamin d supplementation in human participants compared to placebo or no treatment.
When a full protocol file is available, it can be found at /supplements/vitamin-d/protocol/.
Search Strategy
Databases searched: PubMed, Cochrane, Google Scholar
Last searched: 2026-02-22T03:00:00Z
Studies reviewed: 390
Studies meeting inclusion criteria: 242
Searches targeted published systematic reviews and meta-analyses of RCTs for each health claim. Individual RCTs were included when no pooled analysis existed.
Study Selection
Each claim was evaluated independently. The PRISMA flow below summarizes the selection process per outcome.
| Claim | Identified | Screened | Excluded | Included |
|---|---|---|---|---|
| Strengthens bones and prevents fractures | 2364 | 1240 | 1085 | 81 |
| Boosts immune function | 3862 | 1870 | 1690 | 43 |
| Reduces depression | 1540 | 680 | 610 | 25 |
| Improves muscle strength | 890 | 420 | 370 | 13 |
| Reduces respiratory infections | 1528 | 1528 | 43 | |
| Reduces cancer mortality | 3664 | 14 | ||
| Prevents type 2 diabetes in prediabetes | 11 | |||
| Improves blood sugar control in type 2 diabetes | 39 | |||
| Reduces risk of falls in older adults | 35 | |||
| Reduces inflammation (CRP and cytokines) | 23 | |||
| Improves sleep quality | 19051 | 3 | ||
| Lowers blood pressure | 8956 | 46 |
Risk of Bias
Assessment tool: Cochrane RoB 2 for RCTs, ROBINS-I for non-randomized studies.
Individual study risk-of-bias assessments are summarized below by claim. Full per-domain assessments will be available in the downloadable study ledger when published.
| Claim | Studies | Low RoB | Some Concerns | High RoB |
|---|---|---|---|---|
| Strengthens bones and prevents fractures | 81 | 3 | 2 | 0 |
| Boosts immune function | 43 | 5 | 0 | 0 |
| Reduces depression | 25 | 1 | 3 | 0 |
| Improves muscle strength | 13 | 3 | 1 | 0 |
| Reduces respiratory infections | 43 | 9 | 1 | 0 |
| Reduces cancer mortality | 14 | 6 | 1 | 0 |
| Prevents type 2 diabetes in prediabetes | 11 | 3 | 3 | 0 |
| Improves blood sugar control in type 2 diabetes | 39 | 2 | 5 | 0 |
| Reduces risk of falls in older adults | 35 | 3 | 0 | 0 |
| Reduces inflammation (CRP and cytokines) | 23 | 1 | 5 | 0 |
| Improves sleep quality | 3 | 2 | 4 | 0 |
| Lowers blood pressure | 46 | 5 | 1 | 0 |
Results
Strengthens bones and prevents fractures
Pooled effect: RR = 0.97 (95% CI: 0.93 to 1.02, p = 0.21)
Heterogeneity: I² = 12.3%, τ² = 0.01, Cochran's Q = 92.4
81 studies with over 53,000 people found vitamin D doesn't significantly reduce fracture risk on its own. The risk ratio of 0.97 means almost no difference from placebo. Combined with calcium, there's a small benefit for hip fractures in elderly people. Bolus (mega-dose) vitamin D may actually increase falls and fractures.
Boosts immune function
Pooled effect: OR = 0.92 (95% CI: 0.86 to 0.99, p = 0.03)
Heterogeneity: I² = 38.2%, τ² = 0.02, Cochran's Q = 69.6
43 studies with nearly 49,000 people found vitamin D reduces the odds of catching acute respiratory infections by about 8%. That's a real but modest effect. The big finding: people who start out very deficient (under 25 nmol/L) get a 37% reduction. Daily dosing works. Bolus dosing doesn't.
Reduces depression
Pooled effect: SMD = -0.28 (95% CI: -0.53 to -0.03, p = 0.03)
Heterogeneity: I² = 71%, τ² = 0.22, Cochran's Q = 82.8
25 studies with about 7,500 people found a small reduction in depression scores with vitamin D. The effect (SMD = -0.28) is small but statistically significant. Bigger effects show up in people with clinical depression and in those who are vitamin D deficient. The prediction interval crosses zero, so not every new study will find a benefit.
Improves muscle strength
Pooled effect: SMD = 0.17 (95% CI: -0.03 to 0.37, p = 0.09)
Heterogeneity: I² = 55%, τ² = 0.05, Cochran's Q = 26.7
13 studies with about 2,200 people found no significant effect of vitamin D on muscle strength. The effect size is tiny (SMD = 0.17) and the confidence interval crosses zero. The only positive results came from studies of severely deficient elderly people. If your vitamin D levels are normal, don't expect strength gains.
Reduces respiratory infections
Pooled effect: OR = 0.92 (95% CI: 0.86 to 0.99, p = 0.018)
Heterogeneity: I² = 35.6%, τ² =
43 trials with nearly 49,000 people found vitamin D reduces respiratory infections by about 8% overall. Daily dosing at 400-1000 IU works best. Bolus (mega-dose) vitamin D doesn't work at all. The biggest benefit goes to people who are very deficient, where daily dosing cuts infection risk by 70%. An earlier gold-standard IPD meta-analysis of 25 trials confirmed these findings.
Reduces cancer mortality
Pooled effect: RR = 0.94 (95% CI: 0.86 to 1.02, p = 0.153)
Heterogeneity: I² = 0%, τ² = 0, Cochran's Q = 10.96
14 trials with over 104,000 people found vitamin D reduces cancer death risk by 6%, but that wasn't quite statistically significant. The key finding: daily dosing reduced cancer mortality by 12% (statistically significant). Bolus dosing did nothing. This is about cancer survival, not cancer prevention. Vitamin D doesn't stop cancer from developing, but daily supplementation appears to help people survive it.
Prevents type 2 diabetes in prediabetes
Pooled effect: RR = 0.9 (95% CI: 0.81 to 0.99, p = 0.05)
Heterogeneity: I² = 0%, τ² =
11 trials with about 5,200 people who had prediabetes found vitamin D reduces the risk of developing type 2 diabetes by about 10%. That's a modest but statistically significant effect. The benefit was stronger in non-obese people (27% reduction) than in obese people (no significant effect). Vitamin D also increased the chance of reverting from prediabetes back to normal blood sugar by 24-48%.
Improves blood sugar control in type 2 diabetes
Pooled effect: WMD = -0.3 (95% CI: -0.43 to -0.18, p = 0.001)
Heterogeneity: I² = %, τ² =
39 trials with about 3,000 people with type 2 diabetes found vitamin D significantly reduces HbA1c by 0.30%, fasting blood sugar by 0.49 mmol/L, and insulin resistance (HOMA-IR by 0.39). These are modest but clinically meaningful improvements. The effects are strongest in people who are vitamin D deficient, overweight, or have poorly controlled diabetes (HbA1c above 8%).
Reduces risk of falls in older adults
Pooled effect: RR = 0.85 (95% CI: 0.74 to 0.95, p = 0.01)
Heterogeneity: I² = 11%, τ² =
35 trials with nearly 59,000 older adults found vitamin D at 800-1000 IU daily reduces fall risk by about 15%. The effect is strongest in people who are deficient, where fall risk drops by 31%. But dose matters a lot. Higher doses (above 1000 IU) don't help and may actually increase fall risk. Annual mega-doses are harmful. One large trial found a 500,000 IU annual dose increased falls by 15%.
Reduces inflammation (CRP and cytokines)
Pooled effect: ES = -0.42 (95% CI: -0.55 to -0.29, p = 0.001)
Heterogeneity: I² = %, τ² =
An umbrella meta-analysis pooling 23 existing meta-analyses found vitamin D significantly reduces CRP (a key inflammation marker) and TNF-alpha. But IL-6 reduction wasn't significant. The biggest catch: the effect depends heavily on who's taking it. People with diabetes see a meaningful CRP drop of 0.67 mg/L. Healthy overweight people see zero effect. The VITAL trial (n=1,054) found only a temporary CRP reduction at year 2 that vanished by year 4.
Improves sleep quality
Pooled effect: WMD = -2.33 (95% CI: -3.09 to -1.57, p = 0.001)
Heterogeneity: I² = 0%, τ² =
A meta-analysis of 3 small trials with about 200 people found vitamin D improved sleep quality scores by 2.33 points on the PSQI scale. That's clinically meaningful (anything above 1.5 points matters). But there's a big problem: all 3 trials came from Iran, used high biweekly doses, and enrolled people with existing sleep or health issues. The largest sleep trial (VITAL-DEP, n=18,353) found no benefit. One well-designed US trial in postmenopausal women found vitamin D actually worsened sleep.
Lowers blood pressure
Pooled effect: WMD = 0 (95% CI: -0.8 to 0.8, p = 0.97)
Heterogeneity: I² = 21%, τ² =
46 trials with over 4,500 people found vitamin D has zero effect on blood pressure. The gold-standard IPD meta-analysis found 0.0 mmHg change in systolic BP and -0.1 mmHg in diastolic BP. An umbrella meta-analysis of 21 meta-analyses found a barely significant 0.69 mmHg reduction, which is clinically meaningless. Observational studies show a strong link between low vitamin D and high blood pressure, but RCTs don't support a causal relationship.
Sensitivity Analysis
Prediction intervals indicate the range of effects expected in a new study. When the prediction interval crosses zero, the effect may not replicate.
| Claim | Effect | 95% PI | Crosses Zero? |
|---|---|---|---|
| Strengthens bones and prevents fractures | 0.97 | 0.88 to 1.07 | No |
| Boosts immune function | 0.92 | 0.76 to 1.11 | No |
| Reduces depression | -0.28 | -1.2 to 0.64 | Yes |
| Improves muscle strength | 0.17 | -0.3 to 0.64 | Yes |
| Reduces respiratory infections | 0.92 | to | No |
| Reduces cancer mortality | 0.94 | to | No |
| Prevents type 2 diabetes in prediabetes | 0.9 | to | No |
| Improves blood sugar control in type 2 diabetes | -0.3 | to | No |
| Reduces risk of falls in older adults | 0.85 | to | No |
| Reduces inflammation (CRP and cytokines) | -0.42 | to | No |
| Improves sleep quality | -2.33 | to | No |
| Lowers blood pressure | 0 | to | No |
Publication Bias
Funnel plots and Egger's regression test were used to assess publication bias where 10 or more studies were available.
| Claim | Egger's p | Interpretation | Trim-and-Fill Estimate |
|---|---|---|---|
| Strengthens bones and prevents fractures | 0.42 | no significant asymmetry detected | 0.98 |
| Boosts immune function | 0.31 | no significant asymmetry detected | 0.93 |
| Reduces depression | 0.08 | borderline asymmetry suggesting possible bias | -0.18 |
| Improves muscle strength | 0.22 | no significant asymmetry detected | — |
| Reduces respiratory infections | — | some funnel asymmetry noted, possibly small-trial bias | — |
| Reduces cancer mortality | 0.6 | no significant asymmetry detected | — |
| Prevents type 2 diabetes in prediabetes | — | no significant heterogeneity or publication bias observed | — |
| Improves blood sugar control in type 2 diabetes | — | not explicitly reported | — |
| Reduces risk of falls in older adults | — | not formally reported in network meta-analysis, low heterogeneity suggests consistent findings | — |
| Reduces inflammation (CRP and cytokines) | — | not uniformly assessed across constituent meta-analyses | — |
| Improves sleep quality | — | not assessed, only 3 studies in pooled analysis | — |
| Lowers blood pressure | 0.6 | no significant asymmetry in Zhang 2020 SBP analysis | — |
Certainty of Evidence
Evidence grades follow a simplified GRADE framework: A (high certainty), B (moderate), C (low), D (very low/insufficient).
| Outcome | Grade | Verdict | Key Limitation |
|---|---|---|---|
| Strengthens bones and prevents fractures | B | maybe | This is a very large evidence base. Low heterogeneity (I2 = 12.3%) means studies agree with each other. They agree that … |
| Boosts immune function | A | works | This is an individual participant data meta-analysis, the gold standard. Moderate heterogeneity (I2 = 38.2%) is … |
| Reduces depression | B | maybe | High heterogeneity (I2 = 71%) means results vary a lot between studies. Some show clear benefits, others show nothing. … |
| Improves muscle strength | C | no-evidence | Moderate heterogeneity (I2 = 55%) reflects the split: deficient people sometimes benefit, replete people don't. Overall, … |
| Reduces respiratory infections | A | works | Based on two large meta-analyses (Jolliffe 2021, Martineau 2017 IPD). Moderate heterogeneity (I2 = 35.6%) is acceptable. … |
| Reduces cancer mortality | B | maybe | Based on the largest IPD meta-analysis to date (Kuznia 2023). Zero heterogeneity (I2 = 0%) means all trials agree. The … |
| Prevents type 2 diabetes in prediabetes | B | maybe | Consistent finding across three separate meta-analyses (Zhang 2020, Sim 2024, Tobias 2025). Zero heterogeneity (I2 = 0%) … |
| Improves blood sugar control in type 2 diabetes | B | maybe | Based on Chen 2024, the most comprehensive meta-analysis of glycemic outcomes in T2D. However, a 2026 meta-analysis … |
| Reduces risk of falls in older adults | A | works | Low heterogeneity (I2 = 11%) means studies agree. The dose-response is U-shaped: 800-1000 IU daily is the sweet spot. … |
| Reduces inflammation (CRP and cytokines) | B | maybe | The umbrella meta-analysis (Moslemi 2022) pools diverse populations, which inflates the overall effect. When you look at … |
| Improves sleep quality | B | maybe | Zero heterogeneity (I2 = 0%) sounds impressive but reflects that all 3 pooled studies are remarkably similar (same … |
| Lowers blood pressure | D | no-evidence | This is one of the most definitive null findings in the vitamin D literature. Low heterogeneity (I2 = 21%) means studies … |
Limitations
- Searches were limited to English-language publications. Non-English studies may be missing.
- Study identification and data extraction were assisted by AI tools. All extracted data has been manually verified against source publications.
- Small-study effects may inflate some pooled estimates, particularly for outcomes with fewer than 10 included trials.
- Supplement formulations, dosages, and populations varied across studies. Subgroup analyses were limited by the number of available studies per subgroup.
- Most included studies relied on published meta-analyses as the primary data source. Individual participant data was not available.
Conflicts of Interest & Disclosures
SnakeOilCheck earns commissions from qualifying purchases made through affiliate links on this site. Our meta-analyses are produced independently and are not influenced by affiliate relationships.
All claims are sourced from PubMed-indexed meta-analyses and RCTs. Every assertion includes a specific citation with PMID for independent verification.
AI-assisted research disclosure: Study identification and data extraction were assisted by AI tools. All extracted data has been manually verified against source publications.
Raw Data
Downloadable study ledger files (CSV, JSON) and verification logs will be published as we complete the transition to our new data format. In the meantime, all source meta-analyses are cited in the claim sections above with DOIs for independent verification.
License: CC BY 4.0
How to Cite
SnakeOilCheck. Vitamin D: Systematic Review and Meta-Analysis. snakeoilcheck.com/supplements/vitamin-d/. Updated 2026-02-21 00:00:00 +0000 UTC.