Two regulators now list a rare, sudden form of vision loss as a side effect of semaglutide, and the US labels do not. The European Medicines Agency concluded in June 2025 that NAION — damage to the optic nerve that usually takes the sight of one eye without pain — is a "very rare" side effect of Ozempic, Rybelsus and Wegovy. Britain's regulator followed in February 2026. On the American side, none of the nine current GLP-1 labels mentions the condition. This page sets out what each regulator decided, what the studies actually measured, and the different eye warning that US labels do carry.
Documents read on 9 October 2026: EMA's PRAC highlights of 6 June 2025, the MHRA Drug Safety Update of 5 February 2026, the nine US labels on DailyMed, and the PubMed records of the studies named. This page reports what they state; it does not assess anyone's risk or suggest any change to treatment.
What each regulator has decided
| Regulator | Date | Decision |
|---|---|---|
| European Medicines Agency (PRAC) | 6 June 2025 | NAION added to semaglutide product information as very rare (up to 1 in 10,000); stop semaglutide if NAION is confirmed |
| UK Medicines and Healthcare products Regulatory Agency | 5 February 2026 | NAION "very rarely reported"; urgent referral for sudden vision loss, including partial; stop if confirmed; product information to be updated "in the coming months" |
| US Food and Drug Administration | — | No NAION wording in any current US GLP-1 label (nine labels, effective dates 14 October 2025 to 28 August 2026) |
The MHRA's update adds two details the EMA summary does not. It counted three UK Yellow Card reports suggestive of NAION with semaglutide from its first authorisation in 2018 to 1 August 2025, against an estimated 10.2 million packs dispensed in five years — and it notes that privately prescribed semaglutide may not appear in a patient's medical record, so clinicians should ask. It also says it is reviewing evidence for other GLP-1 drugs; both decisions so far name semaglutide only.
The absence from US labels is a fact about the documents, not a finding that there is no risk. Labels in different jurisdictions are changed by different processes on different timetables; this page found no FDA safety communication on NAION, and the date each label was last revised is in the sources below.
What the studies measured
The question began with one hospital's records and has since been tested in databases a hundred times larger. The relative figures shrink as the studies grow; the absolute figures stay small throughout.
| Study | Design and size | Comparison | Result |
|---|---|---|---|
| Hathaway et al., JAMA Ophthalmology 2024 | One neuro-ophthalmology service; 710 people with type 2 diabetes and 979 with overweight or obesity | Semaglutide vs other diabetes or weight drugs | Hazard ratio 4.28 (diabetes; 17 vs 6 events) and 7.64 (weight; 20 vs 3 events); 36-month cumulative incidence 8.9% vs 1.8% and 6.7% vs 0.8% |
| Tesfaye et al., Diabetes, Obesity and Metabolism 2026 | Three US insurance-claims databases, 482,912 matched pairs with type 2 diabetes | Starting any GLP-1 drug vs starting an SGLT2 inhibitor | Hazard ratio 1.85; absolute rate difference 0.29 per 1,000 person-years; pooled with earlier semaglutide studies, hazard ratio 2.78 |
Three things matter when reading these numbers:
- The first study's population was already selected for eye disease. Everyone in it had been seen by neuro-ophthalmologists, which is why its cumulative incidences — several per cent — are far above anything seen in the general population. Its authors call it an association and say causality needs further study.
- The comparator changes the answer. Semaglutide against "other weight drugs" is a different question from GLP-1 drugs against SGLT2 inhibitors, a class prescribed to similar patients. The active-comparator design gave the smaller ratio.
- Absolute against relative. A hazard ratio of 1.85 sounds large. The same study's rate difference, 0.29 extra cases per 1,000 person-years, works out to about 3 extra cases per 10,000 people treated for a year (glp1ledger's arithmetic) — the same order as the EU's "up to 1 in 10,000" label category.
The literature is still moving: a PubMed search for semaglutide and ischemic optic neuropathy returned 106 records on 9 October 2026. The two studies above are the most cited early report and the largest comparison with an active drug.
The warning US labels do carry: diabetic retinopathy
US labels address a different eye condition. The Ozempic label's section 5.3 reports that in SUSTAIN 6, a two-year cardiovascular trial in type 2 diabetes, diabetic retinopathy complications occurred in 3.0% on semaglutide against 1.8% on placebo — 8.2% against 5.2% in people who already had retinopathy, and 0.7% against 0.4% in those who did not. The label attributes this to the known effect of a rapid fall in blood sugar, which can temporarily worsen retinopathy. Wegovy's label repeats those figures and adds its own diabetes trial: retinopathy reported by 4% on Wegovy and 2.7% on placebo. The Zepbound and Mounjaro labels carry the same caution in general terms, without trial figures. The SUSTAIN 6 numbers are worked through in more detail on Ozempic side effects.
Retinopathy and NAION are different conditions — one affects the blood vessels of the retina in diabetes, the other the blood supply of the optic nerve — and the figures for one say nothing about the other.
The background rate the studies are measured against
NAION occurs without any drug. The MHRA's update gives the published background rates: 7.73 to 11.35 cases per 100,000 person-years in the general population aged 40 and over, and 11.4 to 82 per 100,000 in people with type 2 diabetes, whose condition itself raises the risk; smoking, high blood pressure and high cholesterol are the other risk factors it names. Set beside those figures, the Tesfaye study's rate difference — 0.29 per 1,000, or 29 per 100,000 person-years (glp1ledger's conversion) — is of the same order as the background rate in diabetes, which is why a relative risk near two still means few cases. The overlap between NAION's risk factors and the reasons GLP-1 drugs are prescribed is the central difficulty for every observational study: it is hard to separate the drug from the patient. The other conditions in the class's side-effect record are collected on GLP-1 side effects and the long-term record.
What this page cannot tell you
- Whether semaglutide causes NAION. Both regulators treated the association as strong enough to label; neither study above is a randomised trial, and no randomised trial was designed to measure NAION.
- Whether the risk applies to tirzepatide or other GLP-1 drugs. The regulatory decisions name semaglutide; the larger study grouped all GLP-1 drugs together.
- An individual's risk. That depends on eye and vascular history that no population figure captures.
Sudden loss or rapid worsening of vision is a reason for urgent medical attention, which is what both regulators' advice says; decisions about treatment belong with the prescriber and an eye specialist.
Sources
- European Medicines Agency, "Meeting highlights from the Pharmacovigilance Risk Assessment Committee (PRAC) 2-5 June 2025", 6 June 2025, ema.europa.eu, read 2026-10-09.
- Medicines and Healthcare products Regulatory Agency, "Semaglutide (Wegovy, Ozempic and Rybelsus): risk of Non-arteritic Anterior Ischemic Optic Neuropathy (NAION)", Drug Safety Update, 5 February 2026, gov.uk, read 2026-10-09.
- US prescribing information on DailyMed, manufacturer versions, read 2026-10-09 and searched for "optic neuropathy" and "NAION": OZEMPIC injection (effective 2026-06-01; section 5.3), WEGOVY (2026-06-18; section 5.8), semaglutide tablets (2026-01-30), MOUNJARO (2026-08-27; section 5.7), ZEPBOUND (2026-08-28; section 5.8), TRULICITY (2026-06-16), VICTOZA (2025-10-14), SAXENDA (2026-02-25), FOUNDAYO (2026-07-29).
- Hathaway JT, Shah MP, Hathaway DB, et al., "Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide", JAMA Ophthalmology 2024;142(8):732–739, doi:10.1001/jamaophthalmol.2024.2296, PMID 38958939.
- Tesfaye H, Paik JM, Wexler DJ, et al., "GLP-1RA and the risk of non-arteritic anterior ischaemic optic neuropathy in patients with type 2 diabetes: A population-based study", Diabetes, Obesity and Metabolism 2026;28(2):1517–1528, doi:10.1111/dom.70200, PMID 41104517.
- PubMed search "semaglutide AND (NAION OR ischemic optic neuropathy)", 106 records, run 2026-10-09.