What are the odds of sudden permanent vision loss (NAION) from taking semaglutide (Ozempic/Wegovy)?
Evidence quality 4.75/5
Eight-dimension review score against the quality rubric . Each dimension scored 1–5.
- D1 Source grounding
- 5/5
- D2 Source authority
- 5/5
- D3 Arithmetic
- 4/5
- D4 Uncertainty
- 5/5
- D5 Scope
- 5/5
- D6 Prose
- 5/5
- D7 Perception honesty
- 4/5
- D8 Caveat completeness
- 5/5
≈ As likely as
Perceived
NAION entered public awareness abruptly in mid-2024, when a Harvard referral-clinic study tied semaglutide to a several-fold higher rate of sudden one-eye vision loss, and again in June 2025 when the European Medicines Agency formally added it to the label as a "very rare" side effect. Most people taking these drugs have never heard the term, and the coverage that did reach them swung between "blindness warning" headlines and reassurance that the absolute risk is tiny. The result is a fear that is either absent or wildly miscalibrated, rarely sitting at the actual order of magnitude.
Rough estimate: Perception ranges from unaware to headline-driven 'blindness' alarm; no published survey of NAION awareness or perceived likelihood exists
Source: editorial intuition, not polled
Actual
~1 in 10,000 semaglutide users (EMA 'very rare' frequency band)
adults taking semaglutide (Ozempic/Wegovy/Rybelsus)
Show derivation
Scope is the subgroup of adults taking semaglutide; "lifetime" here is the treatment-course duration, NOT the site-default 59-year adult horizon. EMA's headline "very rare / up to 1 in 10,000" is the EU SmPC frequency-category band (a label tier), not a measured cumulative incidence, so we do not normalize on it. We normalize on the population-based per-person-year rate that EMA and the Simonsen 2025 Danish-Norwegian cohort independently agree on. Simonsen reports an excess of +1.41 NAION events per 10,000 person-years (incidence-rate difference) and a pooled hazard ratio of 2.81; EMA gives "approximately one additional case per 10,000 person-years." Implied diabetic baseline = IRD / (HR - 1) = 1.41 / 1.81 ≈ 0.78 per 10,000 PY; total on-drug rate = baseline × HR ≈ 0.78 × 2.81 ≈ 2.2 per 10,000 PY. Over a central 5-year course: 2.2 × 5 / 10,000 ≈ 1.1e-3 ≈ 1 in 900 (total). The drug-attributable EXCESS over the same course is 1.41 × 5 / 10,000 ≈ 7e-4 ≈ 1 in 1,400. The headline 0.0011 is TOTAL on-drug risk; the excess is reported in caveats. The cumulative figure exceeds the "1 in 10,000" label band precisely because that band is a per-exposure frequency tier while this is cumulative across a multi-year course. The Hathaway 2024 hazard ratios (4.28 in diabetics, 7.64 in overweight patients) are from a neuro-ophthalmology referral cohort and are used only for the association signal, not for absolute risk.
Caveats: The headline 1-in-900 is the TOTAL risk of NAION over a ~5-year course for a dia…
The headline 1-in-900 is the TOTAL risk of NAION over a ~5-year course for a diabetic semaglutide user, reconciled from the population-based Danish-Norwegian rate (Simonsen 2025: +1.41 excess events per 10,000 person-years, pooled HR 2.81) and the EMA's matching "~1 additional case per 10,000 person-years." The drug-attributable EXCESS over the same course is lower, roughly 1 in 1,400. EMA's "very rare / up to 1 in 10,000" is a regulatory frequency-category band, not a cumulative incidence, and is reported as the native figure for recognizability only. The Hathaway 2024 hazard ratios (4.28 in diabetics, 7.64 in overweight patients) come from a neuro-ophthalmology referral clinic and overstate the population effect through selection; they are cited for the association, not the absolute rate. NAION is usually permanent and irreversible, typically affects one eye, and the second eye carries elevated but not certain subsequent risk. The risk is conditional on remaining on the drug — it accrues per year of treatment and does not apply to people not taking semaglutide. General-population background NAION incidence is far lower (2.3-10.2 per 100,000/year in adults over 50). Whether the association is causal remains debated; the regulatory and cohort evidence is consistent but observational, not randomized.
Related risks
Other risks on similar themes — for exploring related fears.
GLP-1 anesthesia aspiration
If you take a GLP-1 drug, will retained stomach contents cause you to aspirate during general anesthesia?
Unsafe imported products
What are the odds of being harmed by an unsafe imported consumer product?
Counterfeit medicine
What are the odds of being harmed or killed by a counterfeit or substandard medicine?
COVID-19
What are the odds of dying from COVID-19 over the course of the pandemic and endemic era?
Pick challenger
A diabetic adult taking semaglutide faces roughly a 1-in-900 chance of non-arteritic anterior ischemic optic neuropathy — sudden, usually permanent loss of vision in one eye — over a five-year course of treatment. That figure is reconciled from two sources that arrived at the same place by different routes: a Danish-Norwegian registry of 61,377 users found an excess of 1.41 NAION cases per 10,000 person-years (pooled hazard ratio 2.81), and the European Medicines Agency, reviewing the same body of evidence in June 2025, put it at “approximately one additional case per 10,000 person-years” and a roughly two-fold increase in risk. The drug-attributable share of that 1-in-900 is smaller — about 1 in 1,400 over the same period — because diabetics carry some baseline NAION risk with or without the drug.
The number most people have actually heard is the EMA’s “very rare, up to 1 in 10,000.” That is a regulatory frequency band, not a measured cumulative risk, and it appears here only because it is the figure on the label. Over a multi-year course the cumulative total runs higher than the per-exposure band suggests, which is the ordinary arithmetic of a small annual rate compounded across years on a chronic medication. The discovery study, a 2024 Harvard analysis out of Massachusetts Eye and Ear, reported far larger hazard ratios — 4.28 in diabetics, 7.64 in overweight patients — but those came from a neuro-ophthalmology referral clinic, where patients with eye problems are precisely who shows up. They are good evidence that an association exists and poor evidence of how large the absolute risk is.
The risk is conditional on being on the drug, and it accrues per year of exposure rather than landing in a single moment, which is why this entry is scoped to semaglutide users rather than the general adult population. Background NAION is genuinely rare — 2.3 to 10.2 cases per 100,000 people per year in adults over 50, and well under that across all ages. Diabetes and sleep apnea already raise that baseline before any drug enters the picture. Whether semaglutide causes NAION or merely travels with it remains observationally unsettled: the registry and regulatory evidence point the same direction and the effect is consistent, but none of it is a randomized trial. What can be said cleanly is the order of magnitude — a rare event made a few times less rare, over years, in a population already predisposed.
Related tidbits
Across a 5-year course of semaglutide, the total risk of the rare optic-nerve condition NAION is about ~1 in 900, of which roughly 1 in 1,400 is excess attributable to the drug. The EMA classes it in the "very rare" band, well below the alarm in some headlines.
Claim ledger
Every number below is what each source reported, with the verbatim quote we relied on and how we arrived at our figure. Click any link to verify directly.
-
[1] European Medicines Agency (Pharmacovigilance Risk Assessment Committee) — PRAC concludes eye condition NAION is a very rare side effect of semaglutide medicines Ozempic, Rybelsus and Wegovy
PRAC concludes eye condition NAION is a very rare side effect of semaglutide medicines Ozempic, Rybelsus and Wegovy- Statistic
NAION classified a 'very rare' side effect of semaglutide (may affect up to 1 in 10,000 people taking it); ~2-fold increased risk in type 2 diabetes; ~1 additional case per 10,000 person-years of treatment- Excerpt
“"NAION is a very rare side effect of semaglutide (meaning it may affect up to 1 in 10,000 people taking semaglutide). [...] Exposure to semaglutide in adults with type 2 diabetes is associated with an approximately two-fold increase in the risk of developing NAION compared with people not taking the medicine. This corresponds to approximately one additional case of NAION per 10,000 person-years of treatment; one person-year corresponds to one person taking semaglutide for one year." ”
- Source data from
- 2025-06-06
- Accessed
- 2026-06-12
- Calculation
- EMA's "very rare / 1 in 10,000" is the EU SmPC frequency-category band (a regulatory label tier), not a measured cumulative incidence. The robust quantitative anchor is the per-person-year rate (~1 additional case per 10,000 PY), which matches the Simonsen IRD. The "~2-fold" matches Simonsen's pooled HR 2.81 in order of magnitude.
-
[2] Diabetes, Obesity and Metabolism (Wiley); Simonsen et al. — Use of semaglutide and risk of non-arteritic anterior ischemic optic neuropathy: A Danish-Norwegian cohort study
Use of semaglutide and risk of non-arteritic anterior ischemic optic neuropathy: A Danish-Norwegian cohort study- Statistic
Pooled HR 2.81 (95% CI 1.67-4.75); incidence-rate difference +1.41 (95% CI +0.53 to +2.29) per 10,000 person-years; 61,377 semaglutide users, 32 NAION events- Excerpt
“"the pooled hazard ratio (HR) was 2.81 (95% confidence interval [CI] 1.67-4.75) [...] the incidence rate difference (IRD) was +1.41 (95% CI +0.53 to +2.29) per 10 000 person-years [...] the use of semaglutide for managing type 2 diabetes is associated with an increased risk of NAION compared with the use of SGLT-2is. However, the absolute risk remains low." ”
- Source data from
- 2025-03-17
- Accessed
- 2026-06-12
- Calculation
- Active-comparator new-user design (semaglutide vs SGLT-2 inhibitors) on Danish + Norwegian national registries — population-based, so usable for absolute risk. IRD +1.41/10,000 PY is the drug-attributable excess. Combined with HR 2.81, implied diabetic baseline ≈ IRD/(HR-1) = 1.41/1.81 ≈ 0.78 per 10,000 PY; implied total on-drug ≈ 0.78 × 2.81 ≈ 2.2 per 10,000 PY. This is the population anchor for the normalized model. Published Wiley/PubMed version cited, not the medRxiv preprint.
- Independence
- Independent of EMA's review population (different national registries, different analytic team), though EMA's pharmacovigilance conclusion drew on the same broad body of cohort evidence. Treat Simonsen as the primary quantitative cohort and EMA as the regulatory synthesis.
-
[3] JAMA Ophthalmology 2024;142(8):732-739; Hathaway JT et al., Massachusetts Eye and Ear / Harvard Medical School — Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide
Risk of Nonarteritic Anterior Ischemic Optic Neuropathy in Patients Prescribed Semaglutide- Statistic
Diabetes cohort HR 4.28 (95% CI 1.62-11.29), 17 vs 6 NAION events; overweight/obese cohort HR 7.64 (95% CI 2.21-26.36), 20 vs 3 events- Excerpt
“"hazard ratio [HR], 4.28; 95% CI, 1.62-11.29 [...] HR, 7.64; 95% CI, 2.21-26.36 [...] 20 NAION events occurred in the prescribed semaglutide cohort vs 3 in the non-GLP-1 RA cohort." ”
- Source data from
- 2024-08-01
- Accessed
- 2026-06-12
- Calculation
- Neuro-ophthalmology referral-clinic cohort (710 diabetic + 979 overweight patients) — the discovery study that first flagged the association. Its absolute event rates are inflated by referral selection and are NOT usable as a base rate; used only for the relative/association signal and as the discovery citation. The larger relative effect vs Simonsen's 2.81 reflects this selection.
-
[4] American Academy of Ophthalmology (EyeNet) — Semaglutide, Weight-Loss Drugs, and Vision Loss (NAION): Current Thinking
Semaglutide, Weight-Loss Drugs, and Vision Loss (NAION): Current Thinking- Statistic
Background NAION incidence 2.3-10.2 per 100,000/yr in adults over 50, 0.54 per 100,000/yr across all ages; type 2 diabetes is a risk factor- Excerpt
“"The condition is rare, with the estimated annual incidence of NAION ranging from 2.3 to 10.2 per 100,000 people in adults over the age of 50, and 0.54 per 100,000 people in all age groups. [...] NAION is more common in individuals with certain health issues, such as type 2 diabetes and sleep apnea." ”
- Source data from
- 2025-10-01
- Accessed
- 2026-06-12
- Calculation
- Supplies the general-population baseline for comparison and a sanity check. The diabetic on-drug rate implied by Simonsen (~2.2 per 10,000 PY = ~22 per 100,000/yr) sits above the general-population over-50 ceiling of 10.2 per 100,000/yr — consistent, since diabetics plus a drug effect should exceed the all-comer over-50 rate.