The 30-second summary
- The question is NAION, a rare form of optic nerve damage that causes painless vision loss, usually in one eye.
- Three large studies, three different answers. A specialist centre found a four to sevenfold risk. A multinational database of nearly 300,000 people found nothing. A third found nothing in year one and roughly double from year two.
- The disagreement is mostly about who was studied, not about who is right, and the honest position is that this is unsettled.
- Diabetic retinopathy is a separate issue and a better understood one: fast falls in blood sugar can worsen existing retinopathy, which is true of insulin too.
- One symptom is an emergency: sudden painless loss of vision in one eye, today.
What NAION is
Nonarteritic anterior ischaemic optic neuropathy is a mouthful for a simple mechanism. The optic nerve carries signal from the eye to the brain, and the front of it has a modest blood supply. When that supply falls short, nerve tissue is damaged, and the damage is usually permanent.
It looks like this: painless loss of vision in one eye, frequently noticed on waking, often affecting the upper or lower half of the visual field rather than blurring everything. There is no redness, no pain on moving the eye, no warning.
It is uncommon. It is also strongly associated with the same things that bring people to a GLP-1 in the first place: type 2 diabetes, high blood pressure, sleep apnoea, high cholesterol. That overlap is the whole reason this question is difficult, and it is worth holding on to before reading a single hazard ratio.
Study one: the signal
In August 2024, researchers at a single academic centre published the study that started this. They searched a registry of patients seen by neuro-ophthalmologists between December 2017 and November 2023, found 16,827 with no history of NAION, and compared those prescribed semaglutide with those prescribed other medicines for the same purpose, matching for sex, age, hypertension, diabetes, sleep apnoea, obesity, cholesterol and coronary disease.
| Group | NAION events | 36-month cumulative incidence | Hazard ratio |
|---|---|---|---|
| Type 2 diabetes, semaglutide | 17 | 8.9% | 4.28 (1.62 to 11.29) |
| Type 2 diabetes, other medicines | 6 | 1.8% | reference |
| Overweight or obese, semaglutide | 20 | 6.7% | 7.64 (2.21 to 26.36) |
| Overweight or obese, other medicines | 3 | 0.8% | reference |
Those are large effects and the authors were careful, concluding only that the findings suggest an association and that causality would need further study.
One number deserves attention before anything else. An 8.9% cumulative incidence of NAION over three years is nothing like the rate in the general population, where the condition is rare. That is not an error. It is a description of who walks into a neuro-ophthalmology clinic at a major academic hospital: people with optic nerve problems. The study is internally valid and the population it describes is not you.
Study two: nothing
In 2025, a retrospective cohort study used a global electronic records database covering 160 healthcare organisations across 21 countries, enrolling from January 2017 to August 2023 with follow-up to August 2024. It split people into three groups and matched them one to one on age, sex, BMI, HbA1c, medications and comorbidities.
- Type 2 diabetes only: 37,314 people
- Obesity only: 129,690 people
- Both: 130,216 people
At one, two and three years, in every one of those groups, semaglutide was not associated with NAION. Some hazard ratios were above one and some below, and all of the confidence intervals crossed it comfortably. The authors concluded that avoiding semaglutide on NAION grounds alone may not be warranted.
Study three: it depends how long
Also in 2025, a third group used a different database, TriNetX, covering 3,344,205 patients with diabetes, and matched 174,584 semaglutide users to 174,584 people on non-GLP-1 diabetes medicines.
Their result splits by time:
| Time from starting | Hazard ratio | Significant |
|---|---|---|
| 1 month | 2.99 (0.31 to 28.75) | no |
| 3 months | 1.33 (0.30 to 5.93) | no |
| 6 months | 1.79 (0.60 to 5.35) | no |
| 1 year | 1.94 (0.93 to 4.02) | no |
| 2 years | 2.39 (1.37 to 4.18) | yes |
| 3 years | 2.44 (1.44 to 4.12) | yes |
| 4 years | 2.05 (1.26 to 3.34) | yes |
Nothing in the first year, then a roughly doubled risk that persists. Risk was also raised in people with diabetes and high blood pressure together.
Why they disagree, and what that leaves you with
The three studies are not really arguing. They looked at different people.
The first drew from a specialist referral clinic, which concentrates exactly the patients most likely to develop optic nerve problems and most likely to have them detected. Selection like that inflates effects and is very hard to adjust away.
The second and third drew from general medical records. Between them they cover hundreds of thousands of matched patients, which is the setting closest to an ordinary person on this medicine. They still disagree with each other, which probably comes down to how each defined its population, how long it followed people and how NAION was coded.
The reasonable summary is: if there is a real effect, it is small in absolute terms and it takes years to show up. The absolute numbers matter more than the ratios here. Doubling a rare risk leaves a risk that is still rare.
None of this is a reason to stop a medicine on your own. It is a reason to know one symptom.
The separate question: diabetic retinopathy
This is older, better understood, and more likely to be relevant to you if you have diabetes.
In SUSTAIN 6, a two-year cardiovascular outcomes trial, semaglutide was associated with a significant increase in diabetic retinopathy complications against placebo. A later analysis across the SUSTAIN programme found no imbalance in SUSTAIN 1 to 5 or the Japanese trials, and concluded that most of the SUSTAIN 6 effect could be attributed to the size and speed of the HbA1c fall in the first 16 weeks, in people who already had retinopathy, had poor control at the start, and were on insulin.
This is a known phenomenon rather than a new one. Early worsening of retinopathy after a rapid improvement in blood sugar is well documented with insulin, and guidance exists for it. The same caution is reasonable here.
In practice: if you have diabetes and existing retinopathy, your retinal screening should be current before a big change in control, and your prescriber should know about the retinopathy before the dose climbs.
What to watch for, and when to act
Go the same day for sudden painless loss of vision in one eye, or a dark or missing patch in your field of view, particularly if you notice it on waking. This is true whether or not you take a GLP-1.
Mention at your next appointment blurred vision that persists beyond the first few weeks, blurring in only one eye, or any change that is getting worse rather than settling.
Expect, and do not worry about mild blurring that comes and goes in the early weeks. The lens of the eye takes on and gives up water as blood sugar moves, and vision follows it. It settles as things stabilise.
Keep your screening current if you have diabetes. This is the single most useful thing on the list, and it has nothing to do with which medicine you are on.
What to do with this
None of this changes what is on the plate, and one thing here is worth connecting: the retinopathy signal is about the speed of change, not the fact of it. A plan built to a sensible rate of loss rather than the fastest one available is the version your eyes, kidneys and muscle all prefer.
Sources
- Hathaway JT, et al. Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide. JAMA Ophthalmol 2024. PubMed
- Association between Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy: A Multinational Population-Based Study. Ophthalmology 2025. PubMed
- Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy Risk Among Patients With Diabetes. JAMA Ophthalmol 2025. PubMed
- Nonarteritic Anterior Ischemic Optic Neuropathy and Glucagon-Like Peptide-1 Agonist Therapies: An Important Rare Issue, Lacking Adequate Data. Ophthalmology 2026. PubMed
- Vilsbøll T, et al. Semaglutide, reduction in glycated haemoglobin and the risk of diabetic retinopathy. Diabetes Obes Metab 2018. PubMed
- Semaglutide and non-arteritic anterior ischaemic optic neuropathy: review and interpretation of reported associations. Acta Ophthalmol 2025. PubMed
Medical disclaimer: Articles in the foodose research library are educational, not medical advice. Follow the instructions from your prescriber and the leaflet in your pack. See our full medical disclaimer.