Ozempic and vision loss: what the NAION research shows

The short version

People taking semaglutide develop NAION, a rare form of optic nerve stroke, at higher rates than people who don’t. How much higher depends on which study you read. Some report roughly double the risk. One single-center study reports 4 to 8 times the risk. The largest multi-database study reports a smaller but still detectable increase. No study has proven semaglutide causes NAION. No randomized trial has ever tested the question directly.

Absolute risk varies a lot depending on the population. In the two population-level studies here, a veteran cohort and a 14-database network study, absolute risk stayed well under 1% of treated patients: 0.29% and 14.5 per 100,000 person-years, respectively. The one study reporting a much higher figure, cumulative incidence up to 8.9% at 36 months, came from a single neuro-ophthalmology referral center, a patient population already selected for eye problems and not representative of typical semaglutide users.

As of this writing, the FDA has not added NAION to the Ozempic label. European and global regulators have taken formal action.

Here’s what the evidence shows, and where it disagrees with itself.

What NAION actually is

NAION stands for non-arteritic anterior ischemic optic neuropathy. It’s a stroke of the optic nerve. Blood flow to the nerve gets interrupted, the tissue is damaged, and because optic nerve tissue doesn’t regenerate, the vision loss is usually permanent. It happens in one eye, comes on suddenly and painlessly, and patients often first notice it after waking up, according to specialist literature on the condition.

This is not diabetic retinopathy, which already has a warning on the Ozempic label tied to the SUSTAIN 6 cardiovascular outcomes trial. It’s also not the more common, usually reversible visual side effects people associate with GLP-1 drugs. NAION is rarer, and it’s a different kind of injury: permanent vision loss, usually in one eye. In most cases that means some vision is retained, but a minority of patients lose nearly all light perception in the affected eye, so “not blindness” understates how severe individual cases can get.

NAION also occurs in the general population independent of any medication, at roughly 0.54 to 10.2 cases per 100,000 people per year depending on age. Some baseline number of cases would show up in any large group of semaglutide users even if the drug did nothing at all. That background rate is part of why isolating a true drug effect is hard.

The studies, and why they disagree

Four major studies have looked at semaglutide and NAION. They land in different places.

The single-center cohort that started the alarm. A 2024 JAMA Ophthalmology study by Hathaway and colleagues screened 16,827 neuro-ophthalmology patients treated between December 2017 and November 2023. Among patients with type 2 diabetes, semaglutide use was associated with a 4.28 times higher hazard of NAION (95% CI 1.62–11.29). Among patients who were overweight or obese, the hazard ratio was higher still: 7.64 (95% CI 2.21–26.36). Cumulative incidence at 36 months reached 8.9% versus 1.8% in the diabetes cohort, and 6.7% versus 0.8% in the overweight/obesity cohort. Those are high numbers, but they come from patients already referred to a neuro-ophthalmology clinic, a group with a much higher baseline rate of optic nerve disease than the general population on semaglutide. This is the study most cited in lawsuits and media coverage. It also has the widest confidence intervals and the smallest patient population of the major studies.

A larger, more rigorous veteran cohort. A target-trial-emulation study in US veterans compared 11,478 semaglutide initiators against 90,883 people starting an SGLT2 inhibitor instead, all with type 2 diabetes on metformin, from 2018 to 2025. The design approximates a randomized trial using observational data. It found a hazard ratio of 2.33 (95% CI 1.54–3.54) for semaglutide, with incidence rates of 123 versus 67 per 100,000 person-years. Absolute risk: 0.29% versus 0.13% over a median follow-up of 2.1 years.

The largest study, and a smaller but real effect. An OHDSI network study spanning 14 real-world databases found incidence rate ratios of 1.32 to 1.50, depending on how strictly NAION was defined. The incidence rate among semaglutide users was 14.5 per 100,000 person-years. This is the largest and most geographically diverse dataset of the four, and it produced the smallest effect estimate of the group. A 32 to 50 percent relative increase, consistent across 14 databases, is a modest but real association, not a null result, though it’s far smaller than the single-center hazard ratios above.

A systematic review that found inconsistency, not consensus. A 2025 systematic review of nine studies found that some large cohort and pharmacovigilance studies showed a modest increased risk. Others, including studies by Abbass, Klonoff, and Chou, found no statistically significant association once confounders were accounted for. The review’s conclusion: absolute risk appears low, and current evidence does not support changing clinical practice. It flagged two structural problems across the field. No randomized trial has ever been run, and studies use inconsistent criteria for diagnosing NAION in the first place.

Put together, the range runs from a modest increase (1.3 to 1.5x) to notably elevated (7.6x in a referral-clinic population), with most population-level estimates clustering around 2x. That’s not a resolved question. It’s a signal researchers are still narrowing down.

Why confounding is a real problem here

Every one of these studies is observational. None is a randomized controlled trial, because no trial has been designed or powered to test NAION as a primary outcome. That matters. Diabetes, hypertension, cardiovascular disease, and rapid improvements in blood sugar control are all independent, well-established risk factors for NAION on their own.

People prescribed semaglutide for diabetes or obesity already carry more of these risk factors than the general population. A study comparing semaglutide users to non-users can struggle to separate the drug’s effect from the effect of the underlying conditions the drug treats.

There’s a proposed biological mechanism, though it remains unconfirmed. Rapid glycemic improvement from GLP-1 drugs, similar to what’s seen with insulin, may trigger early worsening of diabetic eye disease and put mechanical or microvascular stress on the optic nerve, according to the Hathaway study’s discussion. If that’s right, any NAION signal might reflect the metabolic shift the drug causes rather than a direct toxic effect on the nerve itself. That distinction matters for how worried patients should be. It hasn’t been tested in a controlled study.

What regulators have actually done

Here the gap between media coverage and regulatory action is widest.

As of this writing (October 2026), the FDA-approved prescribing information for Ozempic contains no NAION or optic neuropathy warning. It carries only the existing warning about diabetic retinopathy complications observed in SUSTAIN 6. Check the current label on Drugs@FDA directly, since FDA revises labels periodically and this status can change.

Other regulators have gone further. In June 2025, the EMA’s Pharmacovigilance Risk Assessment Committee concluded that NAION is a “very rare” side effect of semaglutide, affecting up to 1 in 10,000 users, and recommended adding it to product labeling. That same week, the WHO issued a global medical product alert flagging the same signal to regulators and health care professionals worldwide, based on case reports in its VigiBase database and the EMA’s review. In 2026, the UK’s MHRA and Australia’s Therapeutic Goods Administration both followed with their own safety communications on semaglutide and NAION.

So: the EMA and WHO have acted. The FDA, as of its latest label, has not. That doesn’t mean the FDA has rejected the signal. It means the US review status is simply different right now, and that could change.

What the lawsuits leave out

Lawsuits over Ozempic and vision loss generally point to the Hathaway cohort study and to FAERS adverse event reports, the FDA’s pharmacovigilance database. Reporting odds ratios in FAERS data for semaglutide and NAION have ranged from roughly 9 to 117 depending on the analysis. Those numbers sound alarming in isolation.

But FAERS reports are not a measure of true incidence. Reports spiked sharply after 2024 media and study coverage of NAION, not before. Researchers call this notoriety bias: once a possible side effect gets attention, doctors and patients report it more, whether or not the underlying rate actually changed. The pattern is a known limitation of adverse-event databases. It’s not proof the risk is overstated, but it does mean raw FAERS numbers can’t be read at face value.

None of this says whether any particular lawsuit will succeed. It describes what the cited science can and can’t support: an observed association in some datasets, a weaker one in others, with no study designed to test causation directly.

The North American Neuro-Ophthalmology Society and the American Academy of Ophthalmology issued a joint consensus statement summarizing where the field stands. Their reading: database studies on GLP-1 drugs and NAION average out to a low increased risk, roughly two-fold, while meta-analytic results range from no increased risk to a moderate increased risk.

Despite that signal, the consensus statement does not recommend routine eye exams before starting a GLP-1 medication solely to screen for NAION risk, because the current evidence doesn’t support the practice. It also doesn’t recommend discontinuing GLP-1 therapy based on current evidence.

If you’re managing side effects more broadly, or wondering what happens metabolically when you stop treatment, see our coverage of weight regain after stopping Ozempic or Wegovy and GLP-1 muscle loss during rapid weight loss.

Frequently Asked Questions

Does Ozempic cause blindness? No study shows semaglutide causes NAION with certainty. Multiple large studies show an association with increased relative risk, from roughly 2x in the largest, most rigorous population-level cohorts up to 4-8x in one single-center referral-clinic study. Absolute risk stays well under 1% in the population-level cohorts: 0.29% in the veteran study, 14.5 per 100,000 person-years in the 14-database network study. The single-center study, drawn from a neuro-ophthalmology referral population that doesn’t generalize to typical patients, reported cumulative incidence up to 8.9% at 36 months. NAION, when it happens, usually causes significant vision loss in one eye rather than full blindness, though a minority of cases are severe enough to approach total loss of light perception in that eye. The FDA has not added a NAION warning to the Ozempic label as of this writing; the EMA and WHO have taken formal action internationally.

What is NAION and how is it different from other GLP-1 vision effects? NAION is a sudden interruption of blood flow to the optic nerve, essentially a stroke of the nerve itself. It’s distinct from diabetic retinopathy complications, which already carry an FDA warning tied to the SUSTAIN 6 trial, and from the more common, usually reversible visual side effects linked to GLP-1 drugs. NAION is also rare in the general population regardless of medication, at 0.54 to 10.2 cases per 100,000 people per year depending on age, so some cases would occur with or without semaglutide.

Why hasn’t the FDA added a NAION warning if studies show increased risk? As of this writing, the FDA’s Ozempic label has no NAION-specific language. The EMA’s PRAC concluded in June 2025 that NAION is a very rare side effect, up to 1 in 10,000 users, and recommended a European label update. The WHO issued a global alert the same month. Regulatory review timelines differ between agencies, and the FDA’s status as of this writing may not reflect its final position. Worth checking the current label again as the record develops.

Are the Ozempic lawsuits based on strong scientific evidence? Lawsuits generally cite the 2024 Hathaway cohort study and FAERS adverse event reports. The broader literature is mixed: the largest study, spanning 14 databases, found a smaller but still detectable effect, and a systematic review found several studies showed no statistically significant association after adjusting for confounders. FAERS reports also show a notoriety bias pattern, spiking after media and study coverage rather than before it. That describes the state of the science the lawsuits draw on, not a prediction about legal outcomes.

Who is most at risk, and what should patients do if they notice vision changes? NANOS and the American Academy of Ophthalmology do not recommend routine eye exams before starting a GLP-1 medication solely to screen for NAION, because current evidence doesn’t support the practice. Anyone who experiences sudden, painless vision loss, or rapidly worsening vision in one eye, should seek prompt medical attention. Per EMA guidance, semaglutide should be stopped if NAION is confirmed, under a physician’s direction. General NAION risk factors, like small optic discs and certain cardiovascular conditions, are well established for the condition overall, but they haven’t been confirmed as specific risk multipliers for semaglutide users.

This article is for informational purposes only and is not medical advice. Talk to your healthcare provider before starting, stopping, or changing any medication.