GLP-1 drugs and NAION: what the eye risk data actually shows

What did @weightdoc actually say?

The creator claimed that semaglutide, the active ingredient in Ozempic and Wegovy, carries a small risk of causing NAION, a rare condition that can result in sudden, permanent vision loss. They cited a European Medicines Agency review and put the risk at "one in 10,000 people." They also said the EMA is recommending that product labeling be updated to reflect this risk, and they speculated that the FDA might follow.

To their credit, @weightdoc framed this carefully. They called it a "very rare side effect," avoided catastrophizing, and told viewers to talk to their doctor. That's about as responsible as TikTok medical commentary gets. The core message is grounded in a real regulatory development, not panic speculation.

Does the science back this up?

Yes, with important caveats about causation versus association. The EMA review was real, and the underlying research it drew from is legitimate. A 2024 study by Hathaway et al. in JAMA Ophthalmology found that semaglutide users had a significantly higher incidence of NAION compared to non-users in a retrospective cohort. Among patients with type 2 diabetes, 8.9% of semaglutide users developed NAION versus 1.8% of non-users over a roughly two-year follow-up.

However, that study was observational. Confounding factors, including pre-existing vascular disease, obesity-related optic nerve crowding, and cardiovascular risk profiles, were not fully controlled. The FDA has not issued a label change as of mid-2025, partly because the causal mechanism remains unclear. The EMA acted more cautiously and proactively, which is consistent with its historical approach to drug labeling. The science suggests a probable signal, not a proven causal relationship. That distinction matters.

What did they get wrong (or right)?

The "one in 10,000" figure needs context. The creator presented this number as a settled estimate, but it comes from the EMA's precautionary language rather than from a precise epidemiological calculation. The Hathaway et al. data actually suggested a much higher relative risk in certain subgroups, particularly people who already had crowded optic discs or prior NAION in one eye. Quoting a single flat number without that nuance is not technically wrong, but it is incomplete.

The creator also mispronounced semaglutide multiple times, calling it "samagletide." That's a minor issue but worth noting for credibility. More substantively, they did not mention that NAION has known baseline risk factors that may interact with semaglutide use, including hypertension, diabetes, sleep apnea, and a small cup-to-disc ratio. People with those conditions may face disproportionate risk, and omitting that context leaves viewers without actionable information.

• Got right: EMA review is real and the label update recommendation is accurate.
• Got right: Described NAION correctly as sudden, painless, and usually monocular.
• Got wrong: Presented the 1-in-10,000 figure as precise when it is a rough regulatory estimate.
• Got wrong: Did not mention key risk factors that modify individual risk.

What should you actually know?

If you are currently taking semaglutide or tirzepatide, do not stop based on a TikTok video. The absolute risk here is low, and for most people the cardiovascular and metabolic benefits of these drugs outweigh this rare risk. But low risk is not zero risk, and certain people need to take this more seriously.

Anyone with a prior history of NAION in one eye, a small optic disc, or poorly controlled hypertension should have an explicit conversation with their prescriber before starting or continuing a GLP-1 receptor agonist. A baseline eye exam is reasonable if any of those risk factors apply. Sudden painless vision loss is a medical emergency regardless of cause. If it happens, patients should go to an emergency room immediately, not wait for a telehealth appointment.

The EMA label update is a standard precautionary measure. It does not mean regulators believe semaglutide is dangerous for most people. It means the signal was strong enough to warrant transparency, which is how drug safety is supposed to work.