Post-marketing surveillance · YMYL
Semaglutide + NAION 2024-2026 — the evidence, one study at a time
In July 2024 a Mass Eye and Ear cohort in JAMA Ophthalmology reported a 4-7× higher hazard of non-arteritic anterior ischemic optic neuropathy (NAION) in semaglutide-treated patients. Two years later the FDA has not amended the label, replication cohorts are split, and FAERS reporting is heavily biased by the media coverage of the original signal. Here is what the evidence stack actually supports.
At-a-glance: the five key data points
| Source | Publication · Year | Design | Direction |
|---|---|---|---|
| Hathaway | JAMA Ophthalmology · 2024 | Single-center cohort | HR 4-7× (positive) |
| Chou (Danish reply) | JAMA Ophthalmology · 2024 | Nationwide replication | Signal smaller · CI crosses 1 |
| Novo Nordisk | Company release · 2024 | Pooled RCT safety review | No signal reported |
| Grzybowski | Ophthalmology & Therapy · 2025 | Narrative review | Signal warrants surveillance |
| FAERS | FDA dashboard · 2025 | Spontaneous reporting | ROR elevated · notoriety-biased |
The evidence, one source at a time
JAMA Ophthalmology · 2024
Hathaway JT et al., "Risk of NAION in patients prescribed semaglutide"
Design — Retrospective cohort at Mass Eye and Ear, n=16,827 patients
Semaglutide prescription associated with roughly 4× (T2D subgroup) to 7× (overweight/obesity subgroup) higher hazard of NAION over 36 months vs matched non-semaglutide controls. Absolute risk remained low (~8.9/100,000 patient-years in obesity cohort).
Bias to watch — Single-center referral population; tertiary care selection likely enriches ocular pathology. Not adjusted for T2D duration.
Read primary sourceJAMA Ophthalmology (2024 comment/reply) · 2024
Chou C-Y et al., "Semaglutide and optic nerve disorders — Danish cohort"
Design — Danish nationwide cohort replication attempt
Did not reproduce the Mass Eye and Ear signal at the same magnitude. Point estimates trended toward higher risk in semaglutide users but confidence intervals crossed unity — signal is smaller than the Hathaway cohort or not present at the population level.
Bias to watch — Nationwide EHR — very high external validity but lower ability to detect specialty-clinic outcomes coded to non-NAION optic-nerve categories.
Read primary sourceNovo Nordisk press release · 2024
Novo Nordisk safety review letter (unpublished; company release)
Design — Internal pooled safety database review
Company reported no increased NAION rate in the pooled clinical-trial database. Statistical power for a rare event (~1-8 cases per 100,000 patient-years) is inherently limited in RCT data.
Bias to watch — Manufacturer-authored, not peer-reviewed. RCTs enroll healthier populations than real-world post-marketing exposure.
Read primary sourceOphthalmology and Therapy · 2025
Grzybowski A et al., "GLP-1 receptor agonists and ophthalmic risks — review"
Design — Narrative review of the 2024-2025 evidence base
Concluded that the NAION signal warrants continued surveillance and prescriber awareness but does not currently meet the causation-inference threshold for label change. Recommended documenting baseline optic-nerve status in high-risk patients (small crowded disc, prior optic neuropathy in the fellow eye).
Bias to watch — Review, not new data — reflects the consensus interpretive view as of early 2025.
Read primary sourceFDA FAERS Public Dashboard · 2025
FDA FAERS disproportionality analysis (public dashboard)
Design — Post-marketing spontaneous adverse-event reports
Reporting Odds Ratio for NAION among semaglutide-associated reports is disproportionately elevated versus the FAERS reference population. Absolute number of reports remains under 200/year — orders of magnitude below common GI events.
Bias to watch — FAERS is spontaneous reporting — subject to notoriety bias (heavily-covered signals get reported more), stimulated reporting (post-JAMA press coverage drove increased reporting), and no denominator (no exposure data).
Read primary source
How to interpret the FAERS numbers
FAERS (FDA Adverse Event Reporting System) is a spontaneous reporting database. Anyone — patient, clinician, manufacturer, attorney — can file a report. It is designed to surface hypotheses, not confirm them. Three biases dominate interpretation of any FAERS signal, and all three are active for semaglutide + NAION:
- Notoriety bias. Once the Hathaway paper hit the general press in July 2024, reporting of NAION on semaglutide surged. Clinicians who saw a case of NAION in a semaglutide patient became more likely to file the report because they knew it was under investigation.
- No denominator. FAERS counts reports, not rates. Without knowing the exposed population, an elevated absolute count cannot be directly translated to a risk estimate.
- Legal reporting. Plaintiffs'-side attorneys funnel adverse-event reports for clients they are considering signing — a real phenomenon that inflates counts for signals under active litigation. MDL-3094 discovery may quantify this.
Frequently asked
What is NAION?
Non-arteritic anterior ischemic optic neuropathy (NAION) is sudden, painless vision loss in one eye caused by an ischemic event in the anterior optic nerve. It is the most common cause of acute optic neuropathy in adults over 50 — background incidence in the general population is 2-10 per 100,000 per year. Risk factors include hypertension, diabetes, small crowded optic-nerve head, sleep apnea, and prior NAION in the fellow eye. There is no proven effective acute treatment.
Has the FDA added an NAION warning to the semaglutide label?
As of mid-2026, no. The FDA-approved Ozempic and Wegovy labels do not include an NAION warning. FDA is monitoring the signal but has not concluded that the evidence meets its causation-inference threshold for a labeling change. Prescribing information may change as replication cohorts and MDL-3094 discovery outputs accumulate.
Should I stop semaglutide because of the NAION risk?
This is a discussion for your prescriber, not a page. The absolute risk implied by the Hathaway cohort was ~8.9 NAION events per 100,000 patient-years in the overweight/obese subgroup — a small increase over the general-population background. For most patients, cardiovascular and metabolic benefits of GLP-1 therapy substantially outweigh the NAION signal. Patients with a small crowded optic-nerve head, prior NAION in the fellow eye, or unexplained transient vision loss should have this conversation with their prescriber before continuing.
What should I do if I get sudden vision loss on semaglutide?
Seek immediate ophthalmology or emergency-department evaluation the same day. Acute vision loss of any cause has a narrow window for diagnostic workup. Notify your prescriber that you are on semaglutide so it enters the differential and the safety record.
Is tirzepatide (Zepbound, Mounjaro) affected?
Not currently — the 2024 signal is specific to semaglutide. Tirzepatide is chemically distinct (dual GLP-1 + GIP receptor agonist), and post-marketing data for tirzepatide is younger and smaller. Continued surveillance is warranted but no comparable published signal exists as of mid-2026.
Sources
- JAMA Ophthalmology — Hathaway et al. (2024)
- FDA FAERS Public Dashboard
- FDA Ozempic label (accessdata.fda.gov)
- FDA Wegovy label (accessdata.fda.gov)
- AAO — NAION patient reference page
- Patient-facing summary lives at /glp1-side-effects/naion-vision-loss — this page is the clinician-facing full evidence review.
Editorial notes
- Freshness: reviewed July 17, 2026. Refresh triggered by any new peer-reviewed replication cohort, FDA label change, or MDL-3094 discovery milestone.
- Not medical advice: this page describes the published evidence. Individual prescribing decisions are for a licensed clinician who has assessed the patient in person.
- Citation: journalists and researchers are welcome to cite — attribution to Glpverdict is the only requirement.