Ozempic and Blindness: What the NAION Evidence Shows, and Where India Stands

Semaglutide has not been shown to cause blindness. What the research has found is an association with a rare optic-nerve condition called non-arteritic anterior ischaemic optic neuropathy, or NAION, and the size of that association has fallen every time it has been measured in a larger and better-controlled population.

Europe's pooled estimate puts the excess at roughly one additional case per 10,000 person-years of treatment. No regulator anywhere has withdrawn a semaglutide product. What follows is where the frightening number came from, why it shrank, what regulators actually did as opposed to what headlines said they did, and what a sudden change in vision warrants in India.

Key Takeaways:

• The headline hazard ratio of 7.64 came from a single neuro-ophthalmology referral clinic, a population structurally more likely to have NAION than the general public.
• As study populations grew, the estimate fell: from sevenfold in one referral clinic to roughly twofold in multinational data, and not statistically significant in several large active-comparator comparisons.
• The EMA (June 2025), MHRA (February 2026) and TGA (July 2026) all added NAION as a "very rare" label event. No regulator has withdrawn or suspended any semaglutide product.
• Blurred vision from glucose shifts, early worsening of existing retinopathy, and NAION are three different events with different causes, timings and prognoses.
• A sudden painless dimming or grey patch in one eye warrants same-day emergency assessment, whatever medicine a person is taking.

The figure that travelled came from one study: Hathaway and colleagues, published in JAMA Ophthalmology in 2024. It was a retrospective, propensity-matched cohort drawn from patients evaluated by neuro-ophthalmologists at a single academic centre in Boston between December 2017 and November 2023, screened from 16,827 patients with no history of NAION.

In its overweight-or-obesity cohort, 361 people prescribed semaglutide were matched against 618 people prescribed other weight-management medicines. Twenty NAION events occurred in the semaglutide group against three in the comparator group, giving a 36-month cumulative incidence of 6.7% versus 0.8% and a hazard ratio of 7.64 with a 95% confidence interval of 2.21 to 26.36.

Three features of that table explain most of what happened next, and they are in the paper itself rather than in any rebuttal of it. The confidence interval runs from 2.21 to 26.36, which is the signature of a small sample rather than a precise estimate. The event counts are twenty against three, so a handful of cases drives the entire ratio.

The population is a neuro-ophthalmology referral clinic, not the public, which is why the cumulative incidence reads 6.7% when population NAION rates sit orders of magnitude lower. That is selection, not risk. The authors themselves concluded that the findings suggest an association and that further study would be required to assess causality.

Litigation coverage has amplified the same figure without those qualifiers. A filing records what a claimant alleges. It does not establish that semaglutide caused an individual case, and it establishes nothing at all about causation at population level.

What each major analysis found as the study populations grew:

• Hathaway 2024, single referral centre: 361 semaglutide users vs 618 on other weight-management medicines; 20 events vs 3. HR 7.64 (95% CI 2.21 to 26.36).
• OHDSI network study, 2025: 37.1 million adults across 14 databases, including 810,390 new semaglutide users, with active comparators. Most active-comparator estimates not statistically significant; self-controlled incidence ratio 1.32 (1.14 to 1.54); observed incidence 8.7 to 14.5 per 100,000 person-years.
• US Veterans target-trial emulation, 2026: 239,333 person-years; semaglutide vs SGLT2 inhibitors. HR 2.33 (1.54 to 3.54); cumulative risk 0.29% vs 0.13% over a median 2.1 years.
• EMA pharmacovigilance review, June 2025: Pooled epidemiological evidence, adults with type 2 diabetes vs non-users. Approximately twofold relative increase, equating to about one additional case per 10,000 person-years.

The trajectory is the finding: Each time the question was asked of a larger, less selected population with a like-for-like comparator, the estimate came down, from sevenfold in a referral clinic to roughly twofold across national and multinational datasets. In the largest network analysis, most comparisons against active alternatives did not reach statistical significance at all, and one large cohort analysis returned a hazard ratio of 1.26 with a confidence interval of 0.94 to 1.70, which includes no effect.

None of that is the same as saying the scare was nonsense. A rare association that survives at roughly double the background rate is exactly what a real signal looks like once it stops being measured in the population most likely to produce it. What the evidence does not support is the claim that semaglutide commonly causes blindness, and heterogeneity between studies plus residual confounding still prevent a causal conclusion.

These are three separate actions taken by three different regulators. No single regulator has done all three, and the third has been done by none.

What each regulator has and has not done:

• NAION added to product information: EMA: June 2025, classified "very rare" (up to 1 in 10,000). MHRA: 5 February 2026, same "very rare" classification. TGA: 23 July 2026, across GLP-1 and dual GIP/GLP-1 products.
• Advice to stop treatment where NAION is confirmed: EMA and MHRA: discontinue where NAION is confirmed. TGA: patients advised not to stop without consulting a doctor. No new prescribing restrictions imposed by any of the three.
• Withdrawal of any product from the market: Not done by any regulator anywhere.

Australia's action was also the broadest in scope, covering the wider GLP-1 and dual-agonist class rather than semaglutide alone, and drawing on 36 NAION reports held on the national adverse-events database at the time. The UK's drug safety update referenced Europe's conclusion and set out the stopping advice, but published no absolute risk figures of its own.

Most accounts that circulate as "the injection damaged my eyes" describe one of three different things, and they behave differently.

How the three different vision events differ:

1. Transient blur from glucose changes: Typical onset: fluctuates as blood glucose moves. Pattern: general blur, commonly in both eyes. Reversibility: often improves as glucose stabilises. What examination finds: refractive or lens change, no optic-disc swelling.
2. Early worsening of existing retinopathy: Typical onset: usually the first weeks or months of rapid glucose improvement, particularly where baseline control was poor. Pattern: depends on the retinal changes present. Reversibility: may improve, but requires retinal assessment. What examination finds: retinal haemorrhage, oedema or vascular change.
3. NAION: Typical onset: sudden, often noticed on waking, usually in one eye. Pattern: one eye, or one section of the visual field. Reversibility: frequently permanent. What examination finds: swollen optic disc, characteristic field loss, often an afferent pupil defect.

Transient blur follows osmotic change in the lens as blood glucose shifts. Separately, a rapid fall in HbA1c can temporarily worsen pre-existing diabetic retinopathy, particularly where baseline control was poor or retinopathy was already advanced. Neither of those is NAION, which is an interruption of blood flow to the front of the optic nerve.

Susceptibility to NAION concentrates in people over 50 and in those with diabetes, hypertension, dyslipidaemia, a smoking history, obstructive sleep apnoea or a small, crowded optic disc. Several of those overlap heavily with the reasons someone is prescribed a GLP-1 in the first place, which is part of why observational studies struggle to separate the medicine from the patient.

The same confounding runs through the wider side-effect evidence, and it is why eligibility is judged against a person's history rather than against the drug alone.

India's position, as of September 2026, is an absence in the public record. No publicly indexed CDSCO advisory on semaglutide and NAION, no confirmed India-wide amendment to Indian semaglutide product information, and no guidance from a major Indian ophthalmology or endocrinology body on this question could be located.

That absence should be read as what it is: nothing has been published, which is not the same as Indian authorities having assessed the risk and set it aside. CDSCO's advisory of 10 March 2026 concerns the promotion of GLP-1 medicines and is not an eye-safety action, so the two should not be read together. The GLP-1 supply rules cover what that advisory did do.

Sudden painless dimming, a fixed grey or dark patch, loss of part of the visual field or a rapid fall in colour perception warrants same-day ophthalmological or emergency assessment, whatever medicine a person is taking. Stopping a prescribed medicine on the strength of a headline is not the response. The prescriber is the person to contact, and European and UK regulators advise discontinuation where NAION has actually been confirmed, which is a clinical decision rather than a reader's.

Anyone with a previous NAION, known diabetic retinopathy, sleep apnoea or significant vascular risk has a reason to raise it with the clinician before starting. Anyone with poorly controlled diabetes or established retinopathy has a reason to ask about an eye review before glucose control improves sharply.

A sudden vision change is one of a handful of symptoms on this class that need same-day assessment rather than the next free appointment.

Starting a GLP-1 and want to understand which symptoms need a clinician rather than a headline? Book a consultation with a Voy clinician.

This article is for general information and education only and is not medical advice, diagnosis, or treatment. GLP-1 and other medications referenced are prescription-only and are appropriate only for certain people under the supervision of a qualified clinician. Do not start, stop, or change any medication based on this article. Please consult a registered medical practitioner about your individual circumstances. Information reflects what was available at the time of review and may change.