A whole supplement aisle rests on one number. AREDS reduced the risk of progression to advanced macular degeneration by about 25%. The National Eye Institute says it, every manufacturer prints it, and it is true.
It is also a relative figure, which is the half that goes missing between the journal and the box.
A quarter off a large risk is a large thing. A quarter off a tiny risk is a rounding error. So the only question that leads anywhere is: 25% of what?
Twenty-eight down to twenty
The original Age-Related Eye Disease Study enrolled 3,640 people aged 55 to 80 across eleven centres, randomised them to antioxidants, zinc, both, or placebo, and followed them for an average of 6.3 years. Only 2.4% were lost along the way, which for a six-year trial is unusually tidy.
Among participants in the two higher-risk categories, the estimated probability of progressing to advanced AMD within five years was 28% on placebo and 20% on antioxidants plus zinc.
Eight percentage points. That is it. That is the number the entire category is built on, and it is a genuine result from a well-run trial, and it is much smaller than the shelf suggests.
The people the eight points do not reach
AREDS sorted its participants by how much disease they already had. Those with extensive small drusen, non-extensive intermediate drusen, or pigment abnormalities had a 1.3% probability of progressing to advanced AMD over five years.
One point three percent.
There is no room in that figure for a supplement to do much of anything. The trial's own authors recalculated the effect with those 1,063 participants removed, and the odds reduction got stronger once they were out. That is the signature of a treatment that only acts on people who have something for it to act on.
AREDS2 did not enrol them at all, and the National Eye Institute is direct about why: the second trial excluded people without AMD or with early AMD because the first trial had shown no benefit for them. At the other end, participants who already had advanced AMD in one eye faced a 43% five-year probability in the fellow eye. Same formulation, wildly different stakes.
The trial the formula is named after did not find what people assume
This is the part that almost never gets printed.
AREDS2 ran from 2006 to 2012 with 4,203 participants aged 50 to 85 and a median of five years of follow-up. It asked whether adding lutein and zeaxanthin, or omega-3 fatty acids, or both, on top of the AREDS formulation, would push progression down further than the AREDS formulation alone.
These are the rates it measured.
| Arm | Progressed by year 5 | Hazard ratio vs placebo |
|---|---|---|
| Placebo | 31% | reference |
| Lutein + zeaxanthin | 29% | 0.90 (98.7% CI 0.76 to 1.07) |
| DHA + EPA (omega-3) | 31% | 0.97 (0.82 to 1.16) |
| Lutein + zeaxanthin and omega-3 | 30% | 0.89 (0.75 to 1.06) |
Kaplan-Meier probabilities, AREDS2 Research Group, JAMA 2013. Every participant was also on the AREDS formulation or a variation of it, so 'placebo' here means the AREDS formula without the additions. Note that all three confidence intervals include 1.
Twenty-nine against thirty-one, with a confidence interval running from 0.76 to 1.07. That interval contains 1, which is the statistical way of saying the trial could not rule out that the difference was chance. The p value was 0.12.
The primary result of AREDS2 was negative.
Addition of lutein + zeaxanthin, DHA + EPA, or both to the AREDS formulation in primary analyses did not further reduce risk of progression to advanced AMD.
So why is the whole aisle named after it?
Because of a safety finding rather than an efficacy one.
The original formula contained beta-carotene, and beta-carotene has a long-standing problem in people who have smoked. Within AREDS2, lung cancers appeared in 23 participants (2.0%) taking beta-carotene against 11 (0.9%) not taking it, concentrated in former smokers. A decade later the odds of having had lung cancer were 1.82 times higher in the group randomised to beta-carotene, while lutein and zeaxanthin showed no comparable signal.
So lutein and zeaxanthin took beta-carotene's place. Not because they beat it in the primary analysis, but because they did not appear to carry the harm.
The bottle marked AREDS 2 is the 2001 formulation with one ingredient exchanged for a safer one. Everything the 25% refers to still comes from the 2001 trial. We wrote about what changed between the two formulas separately.
The ten-year follow-up softens that, a little
In 2022 the researchers went back to 3,882 of the participants. Comparing lutein and zeaxanthin directly against beta-carotene, the hazard ratio for progression to late AMD was 0.85, with a confidence interval of 0.73 to 0.98. That one does not cross 1.
It is a real signal and it supports the swap. It is also an epidemiologic follow-on rather than the randomised comparison the trial was designed around, which makes it weaker evidence. Presenting the two as equivalent is exactly the move this article exists to point out, so we are not going to make it here.
Three things sold as upgrades that the trials tested and rejected
- Omega-3. DHA and EPA added nothing at all: 31% progression with them, 31% without. Products still charge extra for fish oil and still describe it as supporting macular health.
- 80 mg of zinc rather than 25 mg. AREDS2 tested both doses. The hazard ratio for the lower dose against the higher was 1.04, p 0.49. No difference. The bigger number is still printed on the front of boxes as though it were a feature.
- Everything else in the ingredient list. Bilberry, astaxanthin, saffron, vitamin D, omega blends. None of them was in either trial. Whatever their individual merits, none can borrow the 25%, and how to read the panel matters more than any of them.
What 'works' does not mean here
The gap between the claim and the finding
What people hear
- It improves your eyesight
- It treats macular degeneration
- It stops you getting AMD
- You should notice something
What the trials measured
- Whether eyes that already had disease got worse more slowly
- Progression on retinal photographs, not treatment of anything
- Nothing: people without AMD were never enrolled
- Endpoints five and ten years out, with nothing to feel in between
There is one exception worth naming, because it is the outcome people actually care about. In AREDS, the antioxidants-plus-zinc group was the only one with a statistically significant reduction in moderate visual acuity loss, defined as losing fifteen letters or more on a chart. The odds ratio was 0.73. Slowing progression on a photograph turned out to track, modestly, with keeping letters on a chart.
Where that leaves an actual decision
If an eye doctor has told you that you have intermediate AMD in one or both eyes, or advanced AMD in one eye, you are the person both trials were about. Your five-year risk sits somewhere near 28%, or near 43% if the fellow eye already has advanced disease, and taking eight points off that is worth having. Thirteen to one is not a miracle. It is also better odds than most things anyone will offer you.
If nobody has told you that, the trials are silent about you. That is not caution or hedging on our part. Silence is the finding.
And if you are in the group, the only thing carrying the evidence is the formulation at the amounts that were tested: vitamin C 500 mg, vitamin E 400 IU, lutein 10 mg, zeaxanthin 2 mg, zinc 80 mg or 25 mg, copper 2 mg, per day. The name on the box guarantees none of that. The panel on the back does. Ours is a checker that reads it for you.