GLP-1, Ozempic & Vision Loss: The NAION Evidence

Headlines saying “Ozempic causes blindness” are an overstatement of a real but narrow signal. Several observational studies have linked semaglutide (the drug in Ozempic and Wegovy) to a higher relative risk of NAION — non-arteritic anterior ischemic optic neuropathy, a sudden, usually painless loss of vision in one eye. The relative risk is genuine: a 2024 Harvard eye-clinic cohort reported hazard ratios of roughly 4–8 (Hathaway/Rizzo 2024 ^[1]), and larger database studies since have mostly confirmed an elevated signal (Hsu 2025 ^[2]; Simonsen 2025 ^[3]; Tesfaye 2026 ^[4]). But the absolute risk is low and causality is not established — which is why European regulators added NAION as a “very rare” side effect, meaning up to 1 in 10,000 people, not a common one ^[6]. This is also not the everyday transient blurred vision from blood-sugar shifts, and it is a separate question from the diabetic-retinopathy signal. Here is what the evidence actually shows — and the one symptom that means call your eye doctor today.

The honest summary

• NAION is sudden, painless vision loss in one eye. The optic nerve's front portion loses blood flow and the eye wakes up with a missing chunk of vision — often the lower half — usually with no pain. It is the most common acute optic neuropathy in people over 50 and is generally irreversible.
• The relative-risk signal is real in several cohorts. The original 2024 Harvard study found hazard ratios of 4.28 (diabetes) and 7.64 (overweight/obesity) for semaglutide (Hathaway/Rizzo 2024^[1]). Large database and registry studies since have mostly pointed the same direction (Hsu 2025^[2]; Simonsen 2025^[3]; Tesfaye 2026^[4]).
• But the absolute risk is small. In a 482,912-pair claims study the rate difference was about 0.29 extra cases per 1,000 person-years (Tesfaye 2026^[4]); a Danish–Norwegian study found roughly 1.4 extra cases per 10,000 person-years (Simonsen 2025^[3]). EU regulators classified NAION as “very rare” — up to 1 in 10,000 (EMA PRAC 2025^[6]).
• Causality is debated — confounding is a live concern. These are observational studies; diabetes itself damages the optic nerve and microvasculature, and the people prescribed semaglutide are exactly the higher-risk patients. A 78-trial meta-analysis found a NAION signal (Peto OR 3.92) but concluded the data were not yet sufficient to be definitive (Natividade 2025^[5]).
• The broader “blindness” fear is overblown. The same meta-analysis found semaglutide did not raise the risk of general eye disorders or diabetic retinopathy overall (Natividade 2025^[5]). NAION is a specific, rare event — not a general vision-destroying effect.
• One symptom matters most: sudden vision loss in one eye. If part of your vision disappears suddenly — especially the top or bottom half of one eye — treat it as an emergency. Do not wait to see if it passes.

What NAION actually is

NAION — non-arteritic anterior ischemic optic neuropathy — is an interruption of blood supply to the front of the optic nerve (the optic nerve head). The classic presentation is sudden, painless loss of vision in one eye, frequently noticed on waking, often as an altitudinal defect (the upper or, more commonly, the lower half of the field goes dark or grey). It is the most common acute optic neuropathy in adults over 50. It is usually not reversible, there is no proven treatment that restores the lost vision, and a minority of people later develop it in the second eye. Its established risk factors — a small, crowded optic disc (a “disc at risk”), diabetes, high blood pressure, sleep apnea, and overnight blood-pressure dips — overlap heavily with the patients who are prescribed semaglutide, which is central to the causality debate below.

What the 2024 cohort actually found — relative vs absolute risk

The signal started with a retrospective matched-cohort study from the Neuro-Ophthalmology Service at Massachusetts Eye and Ear / Harvard (Hathaway, Rizzo and colleagues), published in JAMA Ophthalmology in 2024 (Hathaway/Rizzo 2024^[1]). It is important to be precise about what it was and was not. It drew on 16,827 patients already seen by neuro-ophthalmologists at one academic eye center — not the general population — so the baseline NAION rate is far higher than in the public. Within that enriched group it compared semaglutide users to people on non-GLP-1 medications. In the type-2-diabetes population, 17 NAION events occurred among semaglutide users versus 6 in the comparator, giving a 36-month cumulative incidence of 8.9% vs 1.8% and a hazard ratio of 4.28 (95% CI 1.62–11.29). In the overweight/obesity population it was 20 events vs 3, a cumulative incidence of 6.7% vs 0.8% and a hazard ratio of 7.64 (95% CI 2.21–26.36). The authors were careful: they wrote the findings “suggest an association” and that, being observational, “future study is required to assess causality.”

The larger follow-up studies — mostly elevated, but small in absolute terms

After the 2024 cohort, much larger database studies tested the signal — and broadly confirmed an elevated relative risk while emphasizing how low the absolute risk is. A TriNetX analysis of 3.3 million patients with diabetes found no significant excess in the first year but an increased risk at 2–4 years (hazard ratios around 2.0–2.4) (Hsu 2025^[2]). A Danish–Norwegian national-registry study (61,000+ semaglutide users, 32 NAION events) found a pooled hazard ratio of 2.81 (95% CI 1.67–4.75) versus SGLT-2 inhibitors — but an absolute rate difference of only about 1.4 extra cases per 10,000 person-years, and the authors stated plainly that “the absolute risk remains low” (Simonsen 2025^[3]). The largest to date, a target-trial-emulation study across three U.S. claims databases with 482,912 propensity-matched pairs, found a hazard ratio of 1.85 (95% CI 1.51–2.27) for GLP-1 agonists versus SGLT-2 inhibitors, with a rate difference of just 0.29 per 1,000 person-years — an 85% relative increase, but, in the authors' words, with “incidence rates and absolute increase in risk… small” (Tesfaye 2026^[4]).

The most balanced read comes from a systematic review and meta-analysis of 78 randomized trials covering 73,640 participants (Natividade 2025^[5]). Two findings matter. First, it reassures on the broad fear: semaglutide did not increase the risk of general eye disorders (OR 1.01) or diabetic retinopathy (OR 1.04). Second, on NAION specifically it found a signal (Peto OR 3.92, 95% CI 1.02–15.02) — but a trial-sequential analysis concluded the sample was not yet large enough to draw a definitive conclusion. In short: the direction is consistent across studies, but the question of cause versus confounding by diabetes and vascular risk is not settled.

What regulators have actually done

In June 2025, after reviewing non-clinical data, clinical trials, post-marketing surveillance and the published literature, the European Medicines Agency's safety committee (PRAC) concluded that NAION is a “very rare” side effect of semaglutide — defined as affecting up to 1 in 10,000 people — and recommended adding it to the product information for Ozempic, Wegovy and Rybelsus, with advice that patients who experience sudden vision loss or rapidly worsening eyesight contact their doctor (EMA PRAC 2025^[6]). This is the key calibration: a regulatory body looked at all the evidence and landed on “very rare,” not “causes blindness.” (We are reporting only the EMA action here because it is the regulatory decision we could verify against the official source on the date of publication.)

This is NOT the common blurry vision — and it is not the retinopathy signal

Two other vision changes get conflated with NAION and shouldn't be. First, many people get transient blurred vision in the first weeks of a GLP-1 simply because rapidly improving blood sugar changes the shape and hydration of the lens — this is usually temporary, affects focus rather than knocking out half of one eye's field, and is not NAION. Second, there is a distinct, well-studied signal about early worsening of diabetic retinopathy when blood sugar drops quickly — a separate mechanism in the retina, not the optic nerve. We cover that in depth in our reviews of GLP-1 and diabetic retinopathy and ophthalmology monitoring on GLP-1s, and in our look at glaucoma, eye pressure and dry eye. NAION is its own, much rarer event — and the only one of these where the warning sign is sudden, profound, one-eye vision loss.

What to do with this information

• Don't stop your GLP-1 out of fear alone. The absolute risk is low and these drugs have major cardiovascular and metabolic benefits. Stopping abruptly because of a headline is not the right move — raise it with your prescriber instead.
• Know the one red flag. Sudden, painless loss of vision in one eye — treat it as an emergency and get seen the same day. That is the symptom this entire literature is about.
• Mention your eye history. If you've had NAION in one eye already, a “crowded” optic disc, sleep apnea, or prior optic-nerve problems, tell your prescriber — it should shape the conversation about benefit vs risk.
• Have the shared-decision talk. For most people the math favors continuing; for someone with prior NAION or stacked optic-nerve risk factors, the calculus may differ. That is a conversation for you and your clinician, informed by the real numbers — not by the word “blindness.”

Bottom line

“Ozempic causes blindness” is a mixture of truth and exaggeration. There is a real, repeatedly observed association between semaglutide and NAION — relative risks roughly 1.8 to 8 across studies (Hathaway/Rizzo 2024^[1]; Simonsen 2025^[3]; Tesfaye 2026^[4]) — but the absolute risk is low (around one extra case per 10,000 person-years), causality is not established because diabetes itself harms the optic nerve, and the broad “eye damage” fear isn't borne out (no increase in retinopathy or general eye disorders; Natividade 2025^[5]). EU regulators landed on “very rare” (EMA PRAC 2025^[6]). The practical takeaway: don't quit a GLP-1 out of fear without talking to your clinician — but if you ever lose vision suddenly in one eye, get emergency eye care that day.

This article is educational and is not medical advice. Every claim above is sourced to a peer-reviewed study indexed in PubMed or to an official regulatory statement, verified against the live source before publication. Discuss your own eye history and GLP-1 risk–benefit with your clinician.