The first study found a risk four to eight times higher

In 2024, eye specialists at one Boston hospital saw a rare kind of optic nerve damage in several patients who were taking semaglutide. They checked their records and reported a risk four to nearly eight times higher.

The condition is non-arteritic anterior ischemic optic neuropathy, NAION for short. Blood flow to the optic nerve drops, and vision is lost suddenly and painlessly, usually in one eye. There is no proven treatment.

Headlines shortened all of that to "blindness."

Two years and more than a dozen larger studies later, part of the finding has held and most of its size has not. The reason is a distinction most coverage skipped: "twice the risk" and "one extra case in 10,000" can describe the same result. The first says how much a rare event multiplies. The second says how many people it reaches.

What the larger studies found

In Denmark, people with type 2 diabetes who started semaglutide had about 2.2 NAION diagnoses per 10,000 a year. People who started a different diabetes medicine had about 1.2.

Europe's regulator read the evidence the same way: roughly double the risk, or about one extra case per 10,000 people a year.

The largest gap comes from a study of about 102,000 US veterans with type 2 diabetes: about 12.3 diagnoses per 10,000 a year against about 6.7, or roughly six extra.

Several large studies, including one across 21 countries, found no significant increase.

Why the first number was so high

The Boston patients came from a specialist clinic for optic nerve disease, where eye problems are the reason people are there. The study's figure worked out to 890 cases per 10,000 over three years. Over the same three years, in whole populations of people with diabetes who took semaglutide, the comparable figure runs from about 7 (Denmark) to about 37 (US veterans).

Who is actually at risk?

Start with who gets NAION when no medicine is involved. It affects about 2 to 10 people per 100,000 a year. In one long-running US record, 80% of patients had high blood pressure, 39% had diabetes and 23% had sleep apnea.

Most also share a feature of anatomy that eye doctors call a "disc at risk": a small, crowded optic nerve head with little room to swell. Australia's regulator now states that people with this anatomy are at highest risk "whether or not they are taking a GLP-1 agonist," and that an optometrist or ophthalmologist can check for it.

That list describes many of the people these medicines are prescribed for. With that in view, the evidence sorts people into three groups.

People with type 2 diabetes. This is where the link appears. The extra risk is small, between one and six cases per 10,000 a year depending on the study. A UK study found it highest in the first six months and in people whose blood sugar fell sharply, which is a clue about timing and not yet an explanation.

People taking semaglutide for weight, without diabetes. Studies so far have not found a clear increase. The data are too thin to rule one out.

People who already carry NAION's own risk factors. A crowded optic disc, sleep apnea and high blood pressure raise the odds with or without the medicine.

For tirzepatide, the evidence is thinner still. Australia's expert committee concluded that the evidence may support a signal for semaglutide, but not for tirzepatide. Retatrutide, a different medicine still in clinical trials, was not looked at in any of these studies.

Is it the medicine, or the disease it treats?

This is the open question, and it matters more than any single number.

Doctors tend to give a newer, stronger medicine to people whose diabetes is harder to control. Harder-to-control diabetes is itself a risk for NAION. A nationwide Swedish study tested this in 2026. When it compared people at a similar stage of diabetes, the gap shrank and was no longer statistically clear.

Randomized trials, which assign the medicine by chance and so avoid that problem, have not shown a clear increase either. They also recorded too few cases to settle it.

Europe added a warning. The FDA has not.

What the lawsuits can and cannot settle

People who say they developed NAION after taking these medicines are suing the two companies that make them, Novo Nordisk and Eli Lilly. In December 2025 a federal panel combined 20 cases before one judge in Pennsylvania and noted nine more that may join. The cases name four medicines, two sold for type 2 diabetes and two for weight management, so the litigation reaches further than the evidence, which finds the link in diabetes. The claims include failure to warn. No court has ruled on any of them.

A failure-to-warn case asks what the companies knew and what they told patients and doctors. That is a different question from whether the medicine causes NAION. The number of cases can keep growing whichever way the science lands.

Real, rare, and still unresolved

Both the panic and the dismissal get this wrong.

The risk is real, and it is rare. Three regulators now list it, and it is clearest in people with type 2 diabetes. What remains unresolved is how much of it the medicine itself causes: the newest large study suggests the disease may explain much of it.

Rare is not the same as trivial. When NAION happens the damage is often permanent. In the US record, about 13% of patients were left able to do no more than count fingers with the affected eye, and about one in five later had NAION in the other eye.