Antioxidant pills: the trials that ended the idea

They did not prevent anything, and two of them did harm. In the best-run trials of antioxidant pills, slightly more people died taking them than not: 12.9% against 10.6%.¹ Beta-carotene and vitamin E accounted for the difference.¹ Nothing in the vitamin C or selenium results made up for it.¹

Where the idea came from

The theory was tidy. Free radicals damage cells. Antioxidant vitamins neutralize free radicals. And people who eat the most fruit and vegetables get less cancer and less heart disease. Each of those links holds on its own. The step that failed is the last one: that a capsule could stand in for the diet.

The diet half of the picture is still there, and it is worth being exact about it. Across studies of what people eat, those taking in the most vitamin C had roughly half the colorectal cancer risk of those taking in the least, a relative risk of 0.55.⁵ For prostate cancer the figure was 0.88.⁵ Then the same review looked at vitamin C taken as a supplement and found no link with prostate cancer at all.⁵

That gap is the whole problem in one line. Diet studies cannot separate a nutrient from the person eating it. Somebody with a high vitamin C intake tends also to smoke less and move more, and no amount of statistical adjustment fully removes that. A trial that assigns the pill by chance does remove it. When those trials arrived, the benefit left with the confounding.

What the prevention trials found

The largest accounting tested beta-carotene, vitamin A, vitamin C, vitamin E and selenium against a placebo or nothing. Some participants stayed on them for as long as 12 years, so this is not a case of trials being too short to show anything.¹ Taken all together, the death rate did not move.¹

Then the reviewers set aside the trials whose design left room for wishful results and looked only at the tightly run ones. Those were the trials where the supplement groups did worse.¹ Split by ingredient, beta-carotene and vitamin E raised the death rate, while vitamin A, vitamin C and selenium left it alone.¹

The US Preventive Services Task Force came at the same question from its own pile of evidence and landed in the same place. Vitamin E did not change deaths, heart attacks or cancer diagnoses in healthy adults.² Beta-carotene, with or without vitamin A, moved things the wrong way: lung cancer risk rose about a fifth, an odds ratio of 1.20, and death from heart disease rose about a tenth.² The single hint of benefit, marginally less cancer among multivitamin takers, arrived with limitations the reviewers flagged themselves.²

Beta-carotene is the case study

Beta-carotene deserves a closer look, because it was the most plausible pill in the cabinet. It is the orange pigment in carrots and leafy greens, the marker that shows up in the blood of people who eat well, and it turns into vitamin A inside the body. If any single compound was going to carry the benefit of a vegetable, this was the candidate.

Two separate reviews of the trials tell the same story. The first found no change in cancer overall, and a clear rise in lung cancer: 19% higher, a relative risk of 1.19.³ The risk sat with smokers, and also with participants on low doses.³

The second looked only at lung cancer, in people who did not have it when they enrolled. Beta-carotene raised the risk 16%, and 21% among smokers and asbestos workers.⁴ It also tested whether bigger doses did more damage than small ones, and found no such relationship, which removes the reassuring reading that only a large dose is a problem.⁴ The reviewers added a blunt observation: most patients would rather take a supplement than quit smoking or change what they eat.⁴

Beta-carotene's risk of harm, across reviewsrelative risk (RR or OR)

Lung cancer (USPSTF, OR)OR 1.20

Lung cancer (Zhang 2023, RR)RR 1.19

Lung cancer (Kordiak 2022, RR)RR 1.16

Cardiovascular mortality (USPSTF, OR)OR 1.10

The one place antioxidants do something

There is a real exception, and it is not prevention. In older adults who already had age-related macular degeneration, the disease that erodes central vision, antioxidant vitamins slowed the progression to its late stage.⁶

How much they helped depended on how advanced the disease already was. Among people at the intermediate stage, roughly 78 in every 1000 were spared a progression they would otherwise have had.⁶ In early disease, where few people were going to progress in the first place, the figure was about four in 1000.⁶ The same review picked up the beta-carotene lung cancer signal again, mostly in former smokers, which is why the eye formula was later rebuilt with lutein and zeaxanthin in its place.⁶

Two things separate that result from the prevention trials. Everyone enrolled already had the disease being measured. And the outcome was whether existing damage got worse, not whether new damage appeared. Antioxidants earned a place in treating one eye condition. Nothing about healthy people taking the same pills to stay healthy follows from it.

Memory, and the last positive signals

The cognitive trials are the closest thing to unfinished business here. Results were mixed, and what benefit there was appeared only after a very long wait: a small gain in overall cognitive scores from beta-carotene after an average of 18 years of treatment, and from vitamin C after 5 to 10 years, with nothing at all at earlier points.⁷ Dementia diagnoses did not budge.⁷

One trial in that collection had been designed to look at prostate cancer, and it found significantly more prostate cancer among the men taking vitamin E.⁷ The reviewers' summary judgment was that no vitamin or mineral strategy meaningfully affects cognitive decline, while noting that these late signals are the only positive ones anybody has found and may deserve a proper test.⁷

What remains unknown

The people in these trials were generally well fed. Nobody has run the equivalent experiment in people who are genuinely short of a vitamin, which is the exception printed on the side of every antioxidant product and the one case where the pills might plausibly do what the label says.

The cognition review points out how little evidence exists on supplements started before the age of 60, and how short and insensitive most of the memory testing was.⁷ The beta-carotene reviewers want a safe upper intake level set, since they found harm without any dose that looked clean.^3,4 And the mortality reviewers made a structural point rather than a nutritional one: antioxidant supplements should be treated as medicinal products and evaluated properly before they are marketed.¹