The Vitamin D Paradox: Correlation Does Not Equal Causation

By Michael Edgar

Vitamin D, as a supplement, has gained notoriety in the last couple decades. Deficiency in the vitamin has been associated with an almost ever-expanding assortment of illnesses, and as a result, an almost knee-jerk reaction for its prescription as a supplement has occurred. Regarding supplementation, it is the closest thing we have to a panacea, at least from a marketing standpoint.

But what is vitamin D? Without getting too much into the physiology, vitamin D3 (cholecalciferol) is a micronutrient consumed through one’s diet from items such as fortified dairy products and fish oils, or synthesized within one’s own skin from 7-dehydrocholesterol through ultraviolet irradiation.(1) Vitamin D produced by 7-dehydrocholesterol is dependent on UV intensity, which differs based on season and latitude.(1) Within the liver, vitamin D is converted through a series of pathways to the active substrate, 1,25(OH)2D3, predominantly present in the kidneys.(1)

Now back to our regularly scheduled panacea, ‘Vitamin D as a Remedy For All Conditions.’ Various observational studies over the years have shown that low vitamin D levels are associated with an increased risk for conditions such as multiple sclerosis, type 1 and type 2 diabetes, cardiovascular disease, colon cancer, breast cancer, autoimmunity, allergies, and infectious disease.(2-3) To show how powerful of an intervention it can be, vitamin D status is also inversely associated with all-cause mortality and may even be beneficial for cancer mortality and respiratory disease mortality.(2-4)

This all seems extremely attractive and enticing, and rightfully so. That being said, simply because vitamin D has high-profile news coverage and enthusiastic promotion, we cannot forget the foundation in which this evidence is based. These miraculous relationships have predominantly been found in observational studies and presented in a way to claim causality, which might lead the undiscriminating observer to agree. (2-4)

For the purpose of this article, we will be diving into the literature surrounding bone health and COVID-19.

Bone Health and Vitamin D

Several different observational studies have shown a relationship between bone health and vitamin D. For example, Mezquita‐Raya et al. 2001 investigated this exact topic with 161 postmenopausal women.(5) All were white, ambulatory and in good health, although diagnosed with osteoporosis.(5) The cut-off point for serum vitamin D status was put to 15 ng/ml (37 nM).(5) Low and behold, they found that vitamin D insufficiency was a common risk factor for osteoporosis, specifically postmenopausal osteoporosis, in healthy, community‐dwelling, ambulatory postmenopausal women.(5)

Although this is the case, when we look at randomized controlled trials (RCTs), a different picture is painted. Bolland et al. 2018 performed a meta-analysis of RCTs, which included 81 RCTs reporting either fracture, falls, or bone mineral density. Despite the large number of trials included, they did not find one significant association between vitamin D and fracture, falls or bone density at any site.(6) To add insult to injury, an umbrella review including several hundred (yes, several hundred) systematic reviews and meta-analyses was performed looking at vitamin D and its association to an extensive list of various outcomes.(7) The authors noted that highly convincing evidence of a clear role of vitamin D does not exist for any outcome investigated, although associations with a selection of outcomes are probable, such as childhood Rickets.(7) Their overall message was that although observational studies have identified various links with several diseases, these have either not been evaluated or replicated in RCTs.(7)

I do want to take a moment to emphasize that vitamin D deficiency, in specific at-risk populations can lead to pathophysiology.(8) For example, Rickets is characterised by impaired mineralisation of the bone at the growth plate and osteoid, predominantly affecting infants and toddlers.(8) There are several forms of Rickets, but in the context of this article, nutritional Rickets (NR), also known as osteomalacia, is considered the most relevant.(8) In this regard, global consensus shows both calcium and vitamin D deficiency as factors towards pathophysiology of NR and appropriate medical referral and management may be necessary.(8)

Just like most topics in research, nuance is necessary and as a result, some perspective is needed. For example, on average, Asian women in England had serum vitamin D concentrations of 24.9 nmol/L in the summer and 16.9 nmol/L in the winter.(9) This is in stark comparison to white women in the same community having values of 62.5 nmol/L in the summer and 39.9 nmol/L in the winter.(9) Additionally, a post-mortem based study found that a substantial proportion of people with serum values less than 25 nmol/L had normal bone histology.(10)

Finally, to add more confusion and humour to the situation, an RCT published in JAMA suggested that high dose vitamin D supplements may even increase the risk of fracture.(11) The study protocol prescribed a bolus dose of 500 000 IU cholecalciferol administered orally to older women in the autumn or winter.(11) This was most likely a statistical anomaly, but it did lead to a response by Osteoporosis Canada refuting the findings.(12) They stated this could have been related to the large bolus dose aspect, and not related to appropriate dosing of vitamin D. It also did not change Osteoporosis Canada’s recommendation for 800-2000IU daily.(12)

These various observations, in conjunction with RCT data suggest that it is difficult to extrapolate a low serum vitamin D status to bone disease, let alone other diseases with any degree of certainty.

The takeaway here is not that vitamin D levels do not matter, or that we should ignore them, but whether it is truly a health problem and more so, how do we define normality? To make a parallel to a previous article, this was my final thought in regard to diastasis recti, and how unclear the distinguishment can be between normal and pathological.

COVID-19 and Vitamin D

Moving our conversation onto something that has gained significant traction on social media and various news outlets in recent days, we will discuss the utility of vitamin D for COVID-19.

Several large observational studies have come out recently purporting a link between vitamin D status and COVID-19 with inevitable sensationalism by those that read the headlines. For example, a large meta-analysis of observational studies included 14 studies (999,179 participants) to assess whether low serum 25-OHD is associated with susceptibility to COVID- 19, severity, and mortality related to COVID-19.(13) They used a cut-off value of 20 to 30ng/mL to categorize individuals as low vitamin D (this raises the point of what determines normality).(13) The meta-analysis showed that low serum 25-OHD was associated with higher rates, severity, and mortality from COVID-19 infection.(13)

Some interesting caveats should be noted for this meta-analysis though. For example, age significantly increased the association, yet elderly populations produce roughly 75% less cutaneous vitamin D3 than young adults, making them more prone to lower levels for secondary reasons.(13) Additionally, a funnel plot analysis was performed and found an asymmetrical trend for mortality, severity, and susceptibility.(13) This indicates that publication bias had been involved in the studies included, meaning observational studies that did not show a significant association between COVID-19 and vitamin D were more likely to not be accepted for publication. Despite this, the funnel plot was not shown in the manuscript, so the extent of such bias cannot be commented on.

The authors do note several other possible explanations for the findings. For instance, patients with severe illness are often bedridden and consequently have low vitamin D intake and exposure, leading to a low serum 25-OHD level.(13) More so, obesity was shown to be associated with a low serum 25-OHD level, a co-morbid condition linked to COVID-19 prognosis.(13) Finally, they state that well-designed prospective large cohort studies with rigorous statistical analysis and adequate adjustment to covariates are required to be able to demonstrate a causal relationship.(13) All this considered, even the authors state this research should not be used to make definitive claims of vitamin D utility for COVID-19.(13)

In the realm of RCT data and COVID-19, the data is sparse but there are several studies which have been published investigating this topic. A paper by Murai et al. published in JAMA looked at the effect of a single high dose of vitamin D on hospital length of stay among hospitalized patients with moderate to severe COVID-19. (14) This involved 240 hospitalized patients with a single dose of 200 000 IU of vitamin D, compared with placebo.(14) They found no significant reduction in hospital length of stay and no support for the use of a high dose vitamin D for the treatment of moderate to severe COVID-19 in hospitalized patients.(14)

Just like our JAMA RCT for bolus dosing of vitamin D for bone health, our JAMA RCT on bolus dosing vitamin D for COVID-19 was disputed. (This is why we can’t have nice things). The editorial raised the point that the investigators did not administer Vitamin D until a mean of 10 days from onset of symptoms, and earlier Vitamin D administration could have shortened the hospital stay.(15) The investigators responded that earlier symptom intervention should be performed in a primary health care unit, which was not their objective.(15) This highlights the ongoing back-and-forth within research, especially for highly relevant topics such as COVID-19. Either way, the point remains that insufficient data is available to rule in or out a therapeutic role for Vitamin D for COVID-19.

For current consensus, the National Institute for Health and Care Excellence (NICE), Public Health England, and the Scientific Advisory Committee on Nutrition rapid guideline conclude that there is little good evidence on vitamin D and COVID-19, as no trials exist for vitamin D and prevention.(16) They do maintain support for existing government advice to take 10 μg (400 IU) a day between October and March, to optimise musculoskeletal health.(16)

Two further RCTs were not included in the NICE review that highlight conflicting findings. Firstly, Castillo et al. 2020 showed severity of COVID may be reduced with vitamin D prescription upon hospitalization.(17) This sounds great but there were severely several fundamental problems with this study. Firstly, it was a pilot study, meaning it was not powered to test this hypothesis or answer this question. (Read more about study power here) A pilot study is also typically run as a feasibility study to see if a larger trial can be conducted, as such it asks different questions and any causal relationship found is moot. Additionally, no placebo control was used so we can't really demonstrate a true between-group difference for vitamin D intervention versus no intervention.

Secondly, a smaller RCT from India investigated short-term administration of high-dose vitamin D (1500 μg daily) in asymptomatic or mildly symptomatic COVID-19 patients with vitamin D deficiency. (18) The main outcome measure involved the proportion of patients with COVID-19 negative before day-21.(18) It was found that patients with mild or asymptomatic COVID-19 were more likely to test negative at 21 days following the intervention.(18) Although, the mean duration to COVID-19 negativity was 17.6±6.1 days compared to 17.6±6.4 days in the intervention and control arm, respectively, showing no significant difference to time of recovery.(18) The study also failed to describe any progression of clinical symptoms and lacks adjustment for potential confounders, such as co-morbid conditions. The analyses also did not have any hard endpoints, such as mortality or requirement for more intensive care.

What could explain the discrepancy in findings for these RCTs? There are explanations but differences in participants, type of vitamin D, dose, initial vitamin status, duration of vitamin D supplementation, study endpoints, and risk of bias, to name a few, make consolidated interpretation of these RCTs difficult. As the title of this article highlights, 'correlation does not equal causation’, an aspect of epidemiology frequently overlooked by those advocating for vitamin D supplementation as a panacea, let alone an intervention for COVID-19.

Summary & Takeaways

There are a range of interpretation difficulties with observational studies.

1. Confounding factors - low physical activity may cause low vitamin D, and co-morbid conditions (obesity, diabetes, bone density loss)

2. Classification bias – how vitamin D status is defined (diet versus blood concentrations)

a. Blood concentrations are poorly correlated with nutritional intake

3. Reverse causality - disease may cause reduced exposure to sun

4. Assay method variability – Basic science studies typically use the active form of vitamin D (calcitriol - concentrations tightly regulated in vivo)

a. Conflate organ function health markers with pharmacological action of supplementation

5. Publication bias - reduced publication of negative findings which bias the literature toward positive effects

Vitamin D and its clinical importance.

  1. Vitamin D physiology is complex

  2. Vitamin D is a negative acute phase reactant - It decreases in the setting of other pathological issues, such as obesity or other metabolic syndromes

  3. Distinguishing a pathological range for vitamin D is hard

  4. If vitamin D range is considered pathological, investigate to find an underlying pathology

  5. At accepted therapeutic dosing, it is relatively safe – you can use it at moderate doses for general musculoskeletal wellbeing but understand the literature for this is weak

  6. Avoid false dichotomies - taking vitamin D does not mean avoid vaccinations or other public health measures

  7. Observational studies do not determine causation and correlation does not equal causation

  8. Vitamin D will continue to be sensationalized and haunt us as a panacea for years to come


1. Christakos S, Ajibade DV, Dhawan P, Fechner AJ, Mady LJ. Vitamin D: metabolism. Rheumatic Disease Clinics. 2012 Feb 1;38(1):1-1.

2. Harvey NC, Cooper C. Vitamin D: some perspective please.

3. Holick MF. Vitamin D deficiency. N Engl J Med 2007;357:266-81.

4. Heath AK, Kim IY, Hodge AM, English DR, Muller DC. Vitamin D status and mortality: a systematic review of observational studies. International journal of environmental research and public health. 2019 Jan;16(3):383.

5. Mezquita‐Raya P, Muñoz‐Torres M, De Dios Luna J, Luna V, Lopez‐Rodriguez F, Torres‐Vela E, Escobar‐Jiménez F. Relation between vitamin D insufficiency, bone density, and bone metabolism in healthy postmenopausal women. Journal of Bone and Mineral Research. 2001 Aug;16(8):1408-15.

6. Bolland MJ, Grey A, Avenell A. Effects of vitamin D supplementation on musculoskeletal health: a systematic review, meta-analysis, and trial sequential analysis. The lancet Diabetes & endocrinology. 2018 Nov 1;6(11):847-58.

7. Theodoratou E, Tzoulaki I, Zgaga L, Ioannidis JP. Vitamin D and multiple health outcomes: umbrella review of systematic reviews and meta-analyses of observational studies and randomised trials. Bmj. 2014 Apr 1;348.

8. Siafarikas A, Simm P, Zacharin M, Jefferies C, Lafferty AR, Wheeler BJ, Tham E, Brown J, Biggin A, Hofman P, Woodhead H. Global consensus on nutritional rickets: Implications for Australia. Journal of paediatrics and child health. 2020 Jun;56(6):841-6.

9. Sai AJ, Walters RW, Fang X, Gallagher JC. Relationship between vitamin D, parathyroid hormone, and bone health. J Clin Endocrinol Metab 2011;96:E436-46.

10. Priemel M, von Domarus C, Klatte TO, Kessler S, Schlie J, Meier S, et al. Bone mineralization defects and vitamin D deficiency: histomorphometric analysis of iliac crest bone biopsies and circulating 25-hydroxyvitamin D in 675 patients. J Bone Miner Res 2010;25:305-12.

11. Sanders KM, Stuart AL, Williamson EJ, Simpson JA, Kotowicz MA, Young D, et al. Annual high-dose oral vitamin D and falls and fractures in older women: a randomized controlled trial. JAMA 2010;303:1815-22.

12. Osteoporosis Canada. (2010, May). High-dose oral vitamin D and falls and fractures. Osteoporosis Canada.

13. Akbar MR, Wibowo A, Pranata R, Setiabudiawan B. Low Serum 25-hydroxyvitamin D (Vitamin D) Level Is Associated With Susceptibility to COVID-19, Severity, and Mortality: A Systematic Review and Meta-Analysis. Frontiers in Nutrition. 2021 Mar 29;8:131.

14. Murai IH, Fernandes AL, Sales LP, Pinto AJ, Goessler KF, Duran CS, Silva CB, Franco AS, Macedo MB, Dalmolin HH, Baggio J. Effect of a single high dose of vitamin D3 on hospital length of stay in patients with moderate to severe COVID-19: a randomized clinical trial. Jama. 2021 Mar 16;325(11):1053-60.

15. Vimaleswaran KS, Forouhi NG, Khunti K. Vitamin D and covid-19.

16. Not enough evidence for use of vitamin D in prevention or treatment of COVID-19, NICE rapid guidance concludes. (2020). The Pharmaceutical Journal.

17. Castillo ME, Costa LM, Barrios JM, Díaz JF, Miranda JL, Bouillon R, Gomez JM. Effect of calcifediol treatment and best available therapy versus best available therapy on intensive care unit admission and mortality among patients hospitalized for COVID-19: A pilot randomized clinical study. The Journal of steroid biochemistry and molecular biology. 2020 Oct 1;203:105751.

18. Rastogi A, Bhansali A, Khare N, Suri V, Yaddanapudi N, Sachdeva N, et al. Short term, high-dose vitamin D supplementation for COVID-19 disease: a randomised, placebo-controlled, study (SHADE study). Postgrad Med J. (2020). doi: 10.1136/postgradmedj-2020-139065. [Epub ahead of print].