Reading the evidence

What AREDS2 Actually Tested, And What It Did Not

Three letters and a number appear on half the labels in this category, and the trial behind them is narrower and more interesting than the marketing suggests. It enrolled people who already had an eye disease, it measured whether that disease got worse more slowly, and it took one ingredient out.

Written by Visivra Team at Myvisivra · Reviewed by Dr. Rishi Singh, MD · Updated November 4, 2026 · Reading the evidence

Where the formula came from

Before either trial there was a hypothesis: that oxidative stress contributes to age-related macular degeneration, and that antioxidants plus zinc might slow it. The National Eye Institute set out to test that properly rather than argue about it.

The original AREDS trial tested a specific combination: vitamin C, vitamin E, beta carotene 15 mg, zinc 80 mg as zinc oxide, and copper 2 mg. The copper was there for a specific reason, which is that sustained high zinc intake pushes copper status down over months.

It enrolled people with macular degeneration and asked whether the formula slowed progression to the advanced form. That is the shape of the whole enterprise, and it has been the shape ever since.

A trial of a formula, not of an ingredient

Everything that follows depends on this. AREDS tested a combination at fixed amounts, so the result belongs to the combination. It is not evidence that zinc alone does anything, or that vitamin C alone does, and it is certainly not evidence about a different product sharing two of its ingredient names.

What AREDS2 changed

AREDS2 ran a decade later with two questions rather than one. Would adding lutein and zeaxanthin improve on the original? And could the beta carotene be removed safely?

Two trials, one formula lineage, and one change that matters more than the rest: the beta carotene came out.

The primary AREDS2 report published the main result, and report No. 3 went back to the lutein and zeaxanthin question specifically in a secondary analysis. That distinction between a primary result and a secondary analysis is worth knowing when a label quotes one as though it were the other.

Report 28, the ten-year follow-up, is where the beta carotene question gets its clearest answer: 23 lung cancers in the beta carotene arm against 11 in the arm without it, mostly in former smokers, with an odds ratio of 1.82 (95% CI 1.06–3.12; P=.02).

The population, and why it decides everything

Everyone enrolled in AREDS2 had something wrong with their eyes already. Intermediate age-related macular degeneration, or the advanced form in one eye. The trial measured whether those eyes progressed to late disease.

That single fact governs how the result can honestly be used. It is evidence about slowing something already present, in people who were being followed at an eye clinic, on a specific formula at specific amounts, over five years.

It is not evidence about a person in their fifties with healthy retinas wondering whether to take something. That person is asking a prevention question, and AREDS2 did not enrol them.

Why the distinction is not pedantry

Secondary prevention and primary prevention routinely give different answers in medicine, and there are good reasons for it. A process already underway can be slowed by something that would never have stopped it starting. A population at high risk shows effects a general population does not. And a trial that follows people with an established diagnosis has measurements available that a prevention trial does not.

Five things AREDS2 did not answer

Here is the same point as a table, because these five misreadings are common enough to be worth naming individually.

Five things a label might imply from three words. Only two of them are questions the trial was built to answer.

The first row is the one that matters commercially. A label quoting AREDS2 at a customer with no diagnosis is borrowing a finding from a population that customer is not in, and nothing about the trial supports the transfer.

The prevention question, answered separately

The question most buyers actually have is whether taking these nutrients reduces the chance of ever getting macular degeneration. That has its own evidence base and it is worth reading because the answer is different.

The Cochrane review on prevention asks exactly that, and its conclusions are considerably more cautious than anything on a supplement label. The 2023 Cochrane review covers the other half, slowing progression in people who already have it, which is the AREDS2 territory.

Reading both reviews takes half an hour and it is the single most useful half hour anybody shopping in this category can spend. They are written for clinicians and they are readable by anyone.

• Slowing established disease: the AREDS and AREDS2 territory, with a real result in a specific population.
• Preventing disease starting: a separate literature, with a weaker and more uncertain answer.
• Improving vision in healthy eyes: a third question again, addressed by the contrast-sensitivity trials rather than by AREDS2.
• Restoring vision already lost: not addressed by anything, and not claimed by any reputable product.

What this means when you read a label

Three practical rules follow, and they apply to every product in this category rather than to any particular one.

1. Sharing an ingredient name is not sharing a finding. Lutein on a label and lutein 10 mg a day for five years in a trial are different statements.
2. Check the amounts against the trial amounts. Ten milligrams of lutein and two of zeaxanthin is the pairing the whole category is measured against, and a label that prints no figure cannot be compared at all.
3. Check what the product says about vitamin A. AREDS2 removed beta carotene for a reason, and a formula that puts a vitamin A row back without naming the form has reintroduced the question the trial settled.

On the product this website sells, the second and third of those are open questions. The ingredients page prints the trial amounts beside the label, and the side effects page sets out the vitamin A history in full. Neither is hidden, and neither is flattering.

What the trial is genuinely useful for

None of this is an argument that AREDS2 was a poor trial. It was a large, well-conducted, publicly funded study that answered the question it asked and changed a formula on the basis of a safety signal, which is what good research looks like.

It is genuinely useful in three ways. It establishes that a specific combination slows progression in a specific population, which matters enormously to people in that population. It settled the beta carotene question for everyone. And it set the reference amounts that every carotenoid label in the category is now implicitly measured against.

What it cannot do is carry the weight that supplement marketing puts on it. Three letters and a number on a label are not a finding, and the distance between the two is what this article exists to describe.

References