Supplements and healthspan: what the evidence actually supports

Supplements occupy a strange position in health. Enormous consumer spending, heavy marketing, and a research literature that — for the general population and for hard outcomes — is mostly negative. This guide is about why that gap exists, because the reasoning generalises well beyond supplements.

The recurring pattern

The history of supplement research has a shape that repeats with remarkable consistency:

1. Observational studies find people with higher intake or blood levels of a nutrient have better outcomes.
2. A mechanism is identified that makes this plausible.
3. Supplementation is widely recommended and adopted.
4. Large randomised trials test it against hard endpoints.
5. The trials show no benefit — and occasionally harm.

This has played out for beta-carotene, vitamin E, vitamin C, selenium, B vitamins for cardiovascular prevention, and, in the general population, vitamin D and multivitamins. In several cases the trials found the supplemented group did somewhat worse: trials of high-dose beta-carotene in smokers found increased lung cancer incidence, and some analyses of high-dose vitamin E found increased all-cause mortality.

Large trials have repeatedly failed to show that multivitamins reduce cardiovascular events, cancer, or mortality in generally well-nourished populations. The VITAL trial, testing vitamin D and omega-3 in more than 25,000 participants, did not find reductions in its primary endpoints of invasive cancer or major cardiovascular events.

Why observational promise keeps failing

Three explanations, which the research-reading guide covers in general form.

The healthy-user effect. People who take supplements differ systematically from those who don't — more exercise, better diet, more screening, higher income, more health literacy. Observational studies comparing them are substantially comparing the people, not the pills. This is probably the single largest factor.

Nutrients in food aren't nutrients in isolation. High blood levels of a nutrient may indicate a dietary pattern rich in vegetables, and the pattern may be doing the work. Isolating one compound and delivering it at pharmacological doses is a different intervention from eating the food it came in.

Dose and form matter, often non-monotonically. Physiological amounts from food and high supplemental doses can have opposite effects. Antioxidants are the instructive case: oxidative stress is implicated in disease, but reactive oxygen species also serve signalling functions, including in the adaptive response to exercise. Suppressing them wholesale may interfere with processes you want.

The distinction that explains most of it

The single most useful frame: correcting a deficiency is a different intervention from supplementing an adequate intake.

Where a genuine deficiency exists, supplementation reliably works — this isn't controversial. Vitamin D corrects rickets and osteomalacia. B12 corrects pernicious anaemia. Iron corrects iron-deficiency anaemia. Folate before conception reduces neural tube defects, one of the best-established supplement benefits in medicine. Iodine prevents deficiency disorders.

What repeatedly fails is supplementing people who already have adequate status, in the hope of moving from normal to better-than-normal. The dose-response relationship for most micronutrients is a plateau, not a line: benefit rises steeply as you correct deficiency, then flattens, and for several nutrients turns downward at high doses.

Nearly every negative supplement trial tested the second scenario while the enthusiasm was generated by the first.

Where evidence is more defensible

Some specific cases, with honest characterisation of strength:

• Folic acid before and in early pregnancy. Strong trial evidence for reducing neural tube defects. Among the clearest cases in the field.
• Vitamin D where deficiency is likely. Well established for bone health in deficiency. Evidence for benefit in replete people on non-skeletal outcomes has been largely negative in large trials.
• B12 for vegans and older adults with absorption issues. Deficiency is real, consequential, and easily corrected.
• Iron in diagnosed deficiency. Effective — and worth not taking without diagnosis, since excess iron carries its own risks.
• Creatine for muscle and strength. One of the better-evidenced performance supplements, with reasonable safety data. Effects on healthspan endpoints specifically are unestablished.
• Protein supplementation in older adults with low intake. Reasonable evidence in the context of resistance training and sarcopenia prevention, though food sources work equally well.

What's notable about this list is how much of it is "correct a deficiency" or "support a specific physiological goal," and how little is "extend healthspan in a healthy person." For that last claim, the evidence is essentially absent across the category.

The longevity supplement category

A rapidly growing market of compounds marketed specifically for lifespan or healthspan extension — NAD precursors, resveratrol, various senolytics, metformin for non-diabetics, rapamycin analogues. Some have genuinely interesting mechanistic and animal data behind them.

What essentially none has is randomised human trials with hard endpoints over relevant timescales. Most human data consists of short trials measuring biomarkers — the surrogate endpoint problem, and it's severe here. A compound that raises a blood marker associated with youth in a twelve-week trial has not been shown to extend healthspan; it has been shown to move a number.

Animal lifespan data is worth being specific about. Compounds have extended lifespan in yeast, worms, flies, and mice. Translation from these to humans has a poor track record. Mice are short-lived, genetically homogeneous, kept in controlled conditions, and often studied on diets that make interventions look better than they would in a healthy animal. This isn't a reason to dismiss the research — it's an active and legitimate field — but it is a reason to distinguish "promising preclinical" from "established in humans," and the marketing rarely does.

Our position: the honest summary of the current state is that no supplement has been demonstrated to extend human healthspan. That may change; the research is ongoing and some of it is serious. It hasn't changed yet.

Supplements are not risk-free

Worth stating because "it can't hurt to try" is common and false:

• Demonstrated harms at high doses. Beta-carotene in smokers and high-dose vitamin E are the standard examples, both from large trials.
• Fat-soluble vitamin accumulation. A, D, E, and K are stored rather than excreted, so excess accumulates. Vitamin A toxicity is well documented.
• Drug interactions. St John's wort affects the metabolism of numerous prescription drugs; vitamin K interacts with warfarin; several supplements affect bleeding risk around surgery.
• Regulatory looseness. In many jurisdictions supplements aren't required to demonstrate efficacy before sale, and testing programmes have repeatedly found products with contents differing from their labels, including undeclared pharmaceutical ingredients.
• Opportunity cost. Money and attention spent on supplements aren't spent on the things with strong evidence — sleep, activity, not smoking, blood pressure control.

Why the calculator has no supplement input

Straightforwardly: we could not justify a weight. There is no supplement for which the evidence supports a defensible number of healthy years added or lost in a general population. Including one with a weight near zero would imply we'd measured something we hadn't; giving it a positive weight would misrepresent the literature.

The one genuine exception is deficiency correction, and that isn't a supplement input — it's a medical one, dependent on testing, and outside what a seven-question calculator can assess.

If your interest is healthspan and you're deciding where to put effort, the honest ranking is unambiguous: the factors in this calculator have effect sizes measured in years and evidence spanning decades. Supplements, for people who aren't deficient, have neither. That's not a fashionable conclusion, but it's what the trials show.