By Natalie Howard, DNP, MBA, FNP-C — Board-Certified Family Nurse Practitioner, metabolic medicine practice
No. The eye injury being studied is not blindness, and the distinction matters. Researchers are looking at NAION — non-arteritic anterior ischemic optic neuropathy, a stroke inside the optic nerve that typically takes the upper or lower half of vision in one eye. The largest study to date, published in JAMA Ophthalmology in February 2026, found semaglutide users developed it at roughly twice the rate of a comparison group: 123 versus 67 cases per 100,000 person-years. That is a real signal in a small absolute number — about 1 in 800 people per year.
But that finding comes from observational data, and a 2026 systematic review in Ophthalmology found something most coverage skipped: when researchers separated patients by why they were taking the drug, the increased risk held up in people with diabetes and did not hold up in people taking semaglutide for weight loss.
Watch: Ozempic and Eye Stroke — Who’s Actually at Risk (And Why the Studies Disagree)
What is NAION, and is it the same as going blind?
No. NAION stands for non-arteritic anterior ischemic optic neuropathy. Blood flow to the optic nerve head drops far enough, for long enough, that nerve tissue dies. It is a stroke — it just happens in the optic nerve instead of the brain.
The typical presentation is sudden, painless loss of vision in one eye, most often noticed on waking, and usually altitudinal — meaning the top half or the bottom half of the visual field goes dark rather than the whole eye. Some people recover partial vision. The literature describes the loss as often permanent.
Calling this “blindness” overstates what happens to most patients and understates how serious partial permanent field loss actually is. Neither framing helps you make a decision.
What did the February 2026 JAMA Ophthalmology study find?
Researchers at VA Palo Alto followed 102,361 US veterans with type 2 diabetes, all already taking metformin. Some initiated semaglutide. Some initiated an SGLT2 inhibitor. Over a median of 2.1 years, NAION occurred at 123 cases per 100,000 person-years in the semaglutide group versus 67 per 100,000 in the comparison group — a cumulative risk of 0.29% against 0.13%.
(Heberer K, Bress AP, Cogill S, et al. JAMA Ophthalmol. Published online February 12, 2026. doi:10.1001/jamaophthalmol.2025.6262)
Two things about that comparison are worth holding onto.
First, the doubling is against SGLT2 inhibitors — not against taking nothing. If SGLT2 inhibitors happen to be protective of optic nerve perfusion, and no one has ruled that out, the gap would look identical without semaglutide contributing anything.
Second, this is observational research. The patients were not randomly assigned. Clinicians choose semaglutide for particular patients, and those patients may differ from others in ways the records do not fully capture — what researchers call confounding by indication. Statistical matching reduces that problem. It does not eliminate it.
Why do some studies find a risk and others don’t?
Because the evidence genuinely conflicts, and the pattern of the conflict is informative.
A 2026 systematic review and meta-analysis in Ophthalmology pooled 8 randomized controlled trials covering 31,174 patients with 8 observational studies covering 1,611,278 patients. In the diabetes population, pooled observational data showed 26.7 NAION cases per 100,000 person-years on semaglutide against 18.9 per 100,000 in non-users — hazard ratio 1.85 (95% CI 1.20–2.85), statistically significant.
The randomized trials in that same diabetes population found 5 ischemic optic neuropathy events on semaglutide versus 1 in controls — relative risk 1.76, but with a confidence interval of 0.43 to 7.25.
(Ophthalmology. 2026 May;133(5):589-598. doi:10.1016/j.ophtha.2025.12.022. PMID: 41475544)
That interval deserves a moment. It means the randomized data is equally compatible with semaglutide cutting NAION risk roughly in half and with it multiplying that risk sevenfold. The point estimate leans in the same direction as the observational findings. It simply cannot exclude chance, because NAION is rare enough that a trial of 31,000 people is too small to settle the question.
This is a critical distinction that most coverage collapsed: the randomized trials do not show that the effect is absent. They show that no one has run a study large enough to find out.
The observational studies also disagree with one another — enough that JAMA Ophthalmology published a commentary on the problem titled “Conflicting Results—Need for More Transparent and Reproducible Research” (PMID: 40146117). An earlier TriNetX-based cohort study in the same journal found elevated risk at two, three, and four years but not at one month, six months, or one year (PMID: 40146102).
Does this risk apply to people taking semaglutide for weight loss?
The current evidence does not show that it does.
When the Ophthalmology review stratified its analysis by clinical indication, the weight-loss population showed a hazard ratio of 1.57 with a confidence interval of 0.69 to 3.59 — not statistically significant. The randomized trial data in that group found 4 events versus 1, also not significant.
This is the single most useful finding for most people reading about this, and it received almost no coverage. Same drug. Same review. Different populations, different answer.
There is a straightforward reason to expect that. NAION is strongly age-related and strongly associated with vascular disease. The people studied for diabetes indications had a mean age around 60, were predominantly male, carried a mean BMI near 38, and had diabetes-level vascular risk before any prescription was written. A metabolically healthy 34-year-old taking a GLP-1 for 25 pounds does not resemble that population.
That is not a guarantee of safety. It is an accurate description of what has and has not been measured.
Who is actually at highest risk?
Five factors concentrate NAION risk, independent of any medication:
- Age over 50
- Diabetes, hypertension, or hyperlipidemia
- Obstructive sleep apnea
- A small, crowded optic nerve — clinically described as a “disc at risk”
- Optic nerve head drusen — calcium deposits within the nerve head
The first three describe a very large share of middle-aged adults and do not narrow things much. The last two are the ones that actually sort people, and they are structural: you either have them or you don’t. Neither produces symptoms. Neither is something you would ever notice on your own. Both are visible to an eye care provider during a single dilated exam.
If you are under 50, not diabetic, normotensive, without sleep apnea, and your optic discs are structurally normal, the stratified data supports a very low risk.
Does this apply to Mounjaro or Zepbound?
No data exists either way. Every study discussed here evaluated semaglutide. Mounjaro and Zepbound are tirzepatide, a different molecule with a different receptor profile, and it was not assessed in this literature.
Absence of data is not evidence of safety. It means the question has not been studied.
What should you do if you’re taking a GLP-1?
Nothing in this literature justifies stopping a GLP-1 over eye risk. The cardiovascular and metabolic benefits in these same populations are established at a level of evidence this eye signal does not approach.
Four practical steps:
- Get a baseline dilated eye exam before starting, or early in treatment. This is what identifies a crowded disc or drusen — the two risk factors you cannot otherwise know about.
- Know the symptom. Sudden, painless loss of the upper or lower half of vision in one eye, sometimes with colors appearing washed out. This warrants same-day evaluation, not a scheduled appointment.
- Recheck every 12–18 months if you carry the risk factors above.
- Ask your eye doctor one specific question: “Do I have a crowded disc, or a disc at risk?” It takes them seconds to answer and it moves you off a population statistic and onto your own.
Frequently asked questions
Is the risk of NAION from Ozempic high?
No. Even in the study showing the largest effect, the absolute rate was 123 cases per 100,000 person-years — roughly 1 in 800 per year — compared with 67 per 100,000 in the comparison group. Both numbers are small. The relative doubling and the low absolute risk are both accurate descriptions of the same data.
Can vision loss from NAION be reversed?
The literature describes NAION as a cause of sudden, often permanent vision loss. Some patients recover partial vision. There is no established treatment that reliably restores it, which is why identifying structural risk factors before starting a medication matters more than monitoring for symptoms afterward.
Should I stop taking Ozempic or Wegovy because of this?
That is a decision for you and your prescriber, but nothing in the current evidence supports stopping on this basis alone. The established cardiovascular and metabolic benefits are supported by randomized trial data. The eye signal is drawn primarily from observational data and disappears in the weight-loss population when analyzed separately.
Does the FDA or EMA consider this a confirmed risk?
The European Medicines Agency opened a review of semaglutide and NAION in January 2025. The North American Neuro-Ophthalmology Society and the American Academy of Ophthalmology have issued consensus expert guidance on the association. Regulatory and society positions on this continue to develop; check current labeling for the most recent status.
What is a “disc at risk”?
It describes an optic nerve head with a small scleral canal, where nerve fibers and blood vessels are crowded into a tighter opening with less room to accommodate swelling. It is a structural variant, not a disease, and it is one of the strongest known predisposing factors for NAION. It is identified on a dilated eye exam.
Work with someone who reads the papers
I run Moxie Wellness, a cash-pay telehealth practice focused on metabolic medicine — currently licensed in Florida and Washington, DC. If you’re on a GLP-1 and want a clinician who evaluates the primary literature rather than the headline about it, book a consultation here.
The Health Think Tank is educational and is not a substitute for individualized medical advice. Consult your own clinician about your situation.
Sources
- Heberer K, Bress AP, Cogill S, et al. New-Onset Nonarteritic Anterior Ischemic Optic Neuropathy and Initiators of Semaglutide in US Veterans With Type 2 Diabetes. JAMA Ophthalmol. Published online February 12, 2026. doi:10.1001/jamaophthalmol.2025.6262
- Rate and Risk of Nonarteritic Anterior Ischemic Optic Neuropathy with Semaglutide Use for Diabetes and Weight Loss: A Systematic Review and Meta-Analysis. Ophthalmology. 2026 May;133(5):589-598. doi:10.1016/j.ophtha.2025.12.022. PMID: 41475544
- Semaglutide and Nonarteritic Anterior Ischemic Optic Neuropathy Risk Among Patients With Diabetes. JAMA Ophthalmol. 2025. PMID: 40146102
- Cai CX, Hribar M, Nishimura A. Conflicting Results—Need for More Transparent and Reproducible Research. JAMA Ophthalmol. 2025 May 1;143(5):408-409. doi:10.1001/jamaophthalmol.2025.0350. PMID: 40146117