Vitamin D
The supplement everyone takes and nobody needs to
Description
In 2010, the American supplement industry sold roughly $400 million worth of vitamin D capsules. By 2020, that number had risen to over $1.6 billion. The growth was driven by a particular kind of medical communication that had accumulated across the 2000s: a wave of observational studies, popular books, and authoritative-sounding clinical recommendations suggesting that the modern population was substantially vitamin D deficient and that supplementation could prevent or improve a long list of conditions, including bone disease, cardiovascular disease, certain cancers, autoimmune disorders, depression, and even COVID-19. The supplement became one of the most-tested vitamin levels in primary-care medicine, and supplementation became one of the most-recommended interventions. By 2015, roughly 40 percent of American adults reported taking vitamin D regularly. The recommendation had become, in many practices, a standard piece of preventive-medicine advice.
The randomized controlled trials that the recommendation should have been built on did not exist in 2010. They started reporting results in the second half of the 2010s, and the picture they produced has been substantially different from what the observational studies had suggested. The VITAL trial, published in 2019 — the largest and most rigorous study of vitamin D supplementation ever conducted — randomized 25,871 American adults to either vitamin D plus omega-3 supplementation or placebo and followed them for over five years. The trial found no significant reduction in cardiovascular events, no significant reduction in cancer incidence, no significant reduction in all-cause mortality, and no significant reduction in any of the other major outcomes the supplementation had been recommended for. The D2d trial, published the same year, found no significant reduction in progression to type-2 diabetes in vitamin D-supplemented prediabetic adults. The 2022 meta-analysis pooling these and other trials confirmed the pattern: vitamin D supplementation, in adults who were not severely deficient, produced essentially no measurable health benefits.
The case of vitamin D supplementation has become one of the most-cited recent examples in clinical medicine of how observational research can produce false enthusiasm for an intervention, and of how the corrective process of randomized trials reshapes the evidence base over time. The implications for how patients and clinicians think about preventive supplementation, and for how the supplement industry markets its products, have been substantial. The recommendation that hundreds of millions of people have been taking the supplement on has been largely refuted. The supplement market continues to grow.
The question we’re asking: what did the observational research suggest, what did the randomized trials actually show, and what should the contemporary understanding shape about supplementation practice?
What we’ll see: the observational evidence and the deficiency framing, the VITAL and D2d trials, the meta-analytic consensus, and what survives.
Table of contents
01Observational research and the deficiency framing
The case for widespread vitamin D supplementation accumulated through the 2000s on the basis of observational studies that documented strong correlations between blood levels of vitamin D and a wide range of health outcomes. People with lower vitamin D levels appeared to have higher rates of cardiovascular disease, cancer, autoimmune conditions, depression, fractures, and all-cause mortality. The correlations were consistent across multiple cohort studies in different populations, and they appeared to operate at levels that the medical community considered subclinical — that is, at vitamin D levels above the threshold for the classic deficiency disease, rickets, but below what some researchers were beginning to argue was the optimal range.
The reframing of vitamin D as a substance that adults should be actively supplementing was led by a small number of researchers, including Michael Holick at Boston University Medical Center, whose 2007 book The Vitamin D Solution argued that the modern population was substantially deficient and that supplementation could prevent a long list of conditions. The argument was supported by the observational evidence and by the plausibility of the underlying biology — vitamin D receptors are present in many tissues throughout the body, and the active form of vitamin D affects gene expression in ways that could plausibly produce wide-ranging effects.
02The VITAL trial and what randomization revealed
The VITAL trial was designed to answer the causal question. JoAnn Manson at Harvard Medical School had been advocating for a large randomized trial since the early 2000s, arguing that the observational evidence was suggestive but not definitive and that supplementation recommendations needed a stronger evidence base. The trial received NIH funding in 2010 and began recruiting in late 2011. The design was unusually rigorous. The participants — 25,871 adults aged 50 and older, drawn from across the United States — were randomly assigned to receive either 2,000 IU of vitamin D3 daily, 1 gram of omega-3 daily, both, or placebo. The follow-up averaged 5.3 years. The primary outcomes were invasive cancer and major cardiovascular events.
The results, published in the New England Journal of Medicine in November 2018 and January 2019, were largely negative. Vitamin D supplementation produced no significant reduction in cardiovascular events. It produced no significant reduction in cancer incidence. It produced no significant reduction in all-cause mortality. The trial did show modest reductions in cancer mortality (rather than incidence) in some subgroup analyses, but the primary outcomes were null. The findings were unexpected by the supplementation advocates and were broadly consistent with what the more skeptical members of the field had predicted.
03The meat-analytic consensus and the clinical revisions
The accumulation of randomized trial data through the late 2010s and early 2020s allowed for meta-analyses that pooled the findings across multiple studies and produced more definitive answers about specific outcomes. The 2022 meta-analysis by Bischoff-Ferrari and colleagues, published in JAMA Network Open, pooled data from 22 trials covering over 80,000 participants and found no significant benefit of vitamin D supplementation on cardiovascular outcomes, cancer incidence, fracture incidence in healthy adults, or all-cause mortality. Subgroup analyses found small benefits in specific populations — institutionalized elderly adults at high fracture risk, populations with severe baseline deficiency — but the broad-population recommendations were not supported.
The clinical recommendations have been slowly revised in response. The U.S. Preventive Services Task Force issued updated guidance in 2021 recommending against vitamin D supplementation for the prevention of fractures in healthy postmenopausal women, reversing earlier guidance that had been more permissive. The American Academy of Family Physicians revised its position to recommend against routine vitamin D testing in adults without specific risk factors. The Endocrine Society’s 2024 update to its clinical practice guideline substantially narrowed the population for whom routine vitamin D testing and supplementation was recommended, focusing on populations with specific risk factors rather than broad-based preventive use.
04What survives, and what the case actually shows
The vitamin D case has become one of the most-cited examples in evidence-based medicine of how observational research can produce false enthusiasm for an intervention. The structural problem is well understood. Observational studies cannot distinguish between correlation and causation in the absence of randomization, and the correlations that observational studies identify are often driven by underlying factors that produce both the exposure and the outcome rather than by the exposure causing the outcome. The vitamin D story has played out in this pattern over and over. The strong correlations were real. The causal relationship turned out to be substantially weaker than the correlations suggested.
The lesson for clinical decision-making is specific. The strength of observational correlations should not, by itself, justify confidence in causal relationships. The history of preventive medicine is full of interventions that looked promising in observational data and failed when tested rigorously — hormone replacement therapy, beta-carotene for cancer, vitamin E for cardiovascular disease. Each had observational evidence as strong as vitamin D had. Each was substantially refuted by randomized trials.
05Conclusion
JoAnn Manson and the VITAL trial team continue to publish on the implications of the trial. The clinical revisions have been gradual but real, with major medical societies having issued updated guidance that substantially narrows the population for whom routine supplementation is recommended.