Vision Changes on a GLP-1: What the NAION Evidence Actually Shows

The short version

There is a real safety question about semaglutide and a rare form of sudden vision loss, and it has been answered differently on two continents.

The European regulator concluded in June 2025 that non-arteritic anterior ischemic optic neuropathy — NAION — is a very rare side effect of semaglutide and required it be added to the product information¹. The current US prescribing information for Ozempic, read this month, does not mention optic neuropathy anywhere².

Two 2026 meta-analyses of the same underlying literature reached different conclusions. And the one that split its results by why people were taking the drug found the signal in type 2 diabetes and not in obesity or overweight — which is the indication most readers of this page are treated under³.

First, the thing that is not up for debate: sudden loss of vision, or eyesight that is rapidly getting worse, is an emergency. Contact a clinician the same day, without waiting to see whether it settles¹. Nothing below modifies that.

What NAION is

The optic nerve loses its blood supply at the head, usually overnight, and vision in one eye is painlessly and often permanently reduced on waking. It is uncommon in the general population and its risk factors — diabetes, hypertension, sleep apnea, and a particular optic-disc anatomy — overlap heavily with the people prescribed these drugs. That overlap is the whole difficulty: the population taking semaglutide was already the population that gets NAION.

The two meta-analyses, and why they disagree

Both were published in 2026, both pooled observational studies, and both were graded formally.

The Ophthalmology review is the more conservative read, and it says why. Heterogeneity between studies was extreme — I² above 99% on the odds-ratio analysis — and the result was fragile: dropping one influential study moved the adjusted hazard ratio from 1.63 (non-significant) to 1.92 (95% CI 1.03–2.81, significant)³. A conclusion that flips when you remove one study is not a conclusion you should hold tightly.

The PLoS Medicine review restricted itself to type 2 diabetes and found a consistent doubling — but its most useful sentence is about absolute numbers, not relative ones. Pooled absolute risk was 0.014% (95% CI 0.005% to 0.023%), which the authors translate as roughly one additional case of NAION per 7,000 semaglutide-treated patients per year⁴. The European review put the same thing as approximately one additional case per 10,000 person-years¹.

So: a relative risk that roughly doubles, applied to something that almost never happens.

The part that matters most for this audience

The Ophthalmology meta-analysis was the only one to stratify by treatment indication, and the split is stark³:

• Type 2 diabetes: adjusted HR 1.70 (95% CI 1.10–2.64), low certainty.
• Obesity or overweight: adjusted HR 0.47 (95% CI 0.19–1.13), moderate certainty.

The second estimate does not show increased risk, and it carries higher certainty than the first. The authors are careful not to oversell it — they note the difference may reflect underlying disease-related risk rather than anything about the drug, which is the point³. NAION is a small-vessel event, and people with type 2 diabetes have small-vessel disease. The signal may be tracking the patients rather than the prescription.

Both the European conclusion and every one of the large cohorts behind it rest on adults with type 2 diabetes^1,4,5. The veterans cohort that produced the cleanest single estimate was 85.5% male, mean age 60⁵. A woman in her thirties or forties taking a GLP-1 for weight is not in that data, and her baseline risk of NAION is very much lower before any drug is involved. The earlier multinational population-based study that opened this line of inquiry sits in the same diabetic population⁶.

That is not a reason to dismiss it. It is a reason to know which number applies to you.

The eye warning that IS on the US label

While NAION is absent from the American label, something else is not, and it gets far less attention.

Ozempic's prescribing information carries Warning 5.3, Diabetic Retinopathy Complications. In a two-year trial of patients with type 2 diabetes and high cardiovascular risk, retinopathy complications occurred in 3.0% on semaglutide versus 1.8% on placebo. The absolute increase was concentrated in people who already had retinopathy at baseline — 8.2% versus 5.2% — compared with 0.7% versus 0.4% in those without a known history².

Those are adult figures from the cardiovascular outcomes trial, and the label instructs patients to contact their physician if vision changes during treatment, and clinicians to monitor patients with a history of diabetic retinopathy². The mechanism generally proposed is that rapidly improving glucose control can transiently worsen existing retinopathy — which, if right, means the relevant question is whether you have retinopathy already, not whether you are on this drug.

If you have any diabetic eye disease, that warning is more likely to matter to you than the NAION discussion is.

What to do with this

• Blurred vision in the first weeks, fluctuating with your glucose, is common and usually settles. Mention it; do not panic about it.
• Sudden vision loss in one eye, or vision that is rapidly deteriorating, is a same-day problem. Both the European advice and the label's instruction point the same way: contact a clinician immediately, and if NAION is confirmed, semaglutide should be stopped¹.
• If you have diabetic retinopathy, say so before starting and ask about eye monitoring².
• If you have had NAION in one eye, that belongs in the conversation before a first dose.

Where this sits on the evidence scale

That NAION is a very rare adverse effect of semaglutide in adults with type 2 diabetes: moderate, and regulator-endorsed. Europe reviewed epidemiological studies, trials, post-marketing surveillance and the literature, and set the frequency at very rare — up to 1 in 10,000¹.

That the relative risk roughly doubles in that population: consistent but low-certainty. Two meta-analyses and a large target-trial emulation land near a doubling^4,5, while the most methodologically cautious review calls the pooled evidence fragile and non-significant overall³.

That it applies to women taking a GLP-1 for weight loss: not shown, and the one direct estimate points the other way — adjusted HR 0.47 (0.19–1.13) in the obesity indication, at moderate certainty³.

That the US label warns about it: no. It does not appear². The label's actual eye warning concerns diabetic retinopathy, with real numbers attached.

What would change this page: a cohort of people treated for obesity rather than diabetes, with NAION ascertained by an ophthalmologist rather than a billing code. Every study above relies on diagnostic codes, and all of them say so.