If you saw headlines about a GLP-1 vision loss study this week, take a breath. The study found 3 to 4 extra optic-nerve cases per 10,000 diabetes patients over 18 months. The modeled 18-month risk of a coded optic-nerve event was below 0.1% in both GLP-1 comparisons. These codes did not confirm NAION or measure lifetime risk.
As of October 17, 2026, the day this article was written, October news is revisiting a July paper. An October 15, 2026 ScienceDaily report summarized a July 14, 2026 Rutgers University release, not a new October trial. Both cover a 2026 study in Annals of Internal Medicine published online July 14 and in the September issue.
The paper tracked ischemic optic neuropathy, or ION, which is optic-nerve damage from low blood flow. Doctors focus on nonarteritic anterior ischemic optic neuropathy, or NAION, a sudden, usually painless form. The added risk was small, but sudden vision loss always needs emergency care right away.
What the 2026 GLP-1 vision loss study found
Kamika Reynolds, Chintan Dave, and colleagues studied U.S. insured adults aged 18 to 65 with type 2 diabetes. They used 2017–2022 claims in a target-trial emulation, which analyzes routine records to mimic a trial. The study was funded by the National Institutes of Health. Because treatment was not randomly assigned, researchers used weighting to balance more than 80 measured health factors.
Researchers compared people starting a GLP-1 (glucagon-like peptide-1) drug against two other diabetes pill classes. One group started SGLT2 inhibitors (sodium-glucose cotransporter-2 pills), and the other started DPP4 inhibitors (dipeptidyl peptidase-4 pills). Neither comparator group was untreated, and each adjusted comparison had its own baseline risk.
Over 18 months, modeled coded ION risk was higher in the GLP-1 group against both comparators. Claims lacked a specific NAION code, so broader ION codes served as a proxy. Dave summarized this as "approximately three to four additional cases per 10,000 patients treated with GLP-1s over 18 months."
| 18-month comparison | GLP-1 risk per 10,000 | Comparator risk per 10,000 | Extra cases per 10,000 (95% CI) | Relative ratio (calculated) |
|---|---|---|---|---|
| GLP-1 vs. SGLT2 pills | 8.5 (0.085%) | 5.5 (0.055%) | 3.0 (0.4 to 5.7) | ~1.55x (~55% higher) |
| GLP-1 vs. DPP4 pills | 7.8 (0.078%) | 4.2 (0.042%) | 3.6 (1.1 to 6.1) | ~1.86x (~86% higher) |
The researchers recorded 81 coded ION events among GLP-1 users. Of those 81 cases, 69 (85.2%) occurred above age 50, and 57 (70.3%) occurred in men. Those counts describe cases within this 18-to-65 diabetes cohort, not personal risk or rates over age 65.
What the numbers mean for Ozempic NAION risk
Dividing the rounded 18-month risks gives our calculated ratios of 1.55 (55% higher) and 1.86 (86% higher). Those numbers come from dividing rounded cumulative risks, not paper-reported hazard ratios. Our guide to what relative risk means explains the math, and the absolute gap was 0.030 to 0.036 percentage points.
Earlier studies on semaglutide and vision changes disagreed. A 2024 JAMA Ophthalmology study by Jimena Hathaway and colleagues screened 16,827 records at one referral eye clinic. Over 36 months, semaglutide had a hazard ratio of 4.28 (1.62–11.29) in diabetes and 7.64 (2.21–26.36) in overweight or obesity.
Referral clinics see concentrated eye disease, not general community rates. By contrast, a 2025 multinational study in Ophthalmology by Chien-Chih Chou and colleagues covered 160 healthcare organizations in 21 countries. Across 37,314 diabetes-only, 129,690 obesity-only, and 130,216 combined patients, three-year hazard ratios were 1.51 (0.71–3.25), 0.72 (0.24–2.16), and 1.19 (0.78–1.82).
Every interval in Chou's study crossed 1.0, showing no statistically clear increase without proving zero risk. Meanwhile, a 2026 Swedish nationwide cohort study by Peter Ueda and colleagues compared 107,518 GLP-1 starters with 185,898 SGLT2 starters from 2013 to 2024.
Modeled one-year anterior ION risks in Sweden were 0.04% versus 0.02%. Across the full follow-up, 62 GLP-1 users and 64 SGLT2 users had an event. In an analysis restricted to patients taking metformin at baseline, the one-year risk ratio was 1.40 (95% CI, 0.64–3.05), compared with 1.93 (1.00–3.73) overall. This restriction aimed to reduce confounding by diabetes severity.
An August 2026 Ophthalmology anatomy study analyzed 1,812 unaffected fellow eyes from 2,169 NAION patients alongside both eyes of 465 control patients. However, that study did not test whether GLP-1 drugs independently cause NAION.
What the study does not mean for Mounjaro vision loss
News reports often mention Ozempic, Wegovy, Mounjaro, and Zepbound together, inviting a few common misreadings. First, observational claims cannot prove causation or confirm blindness. The authors noted that "Observed differences may reflect residual confounding," since records lacked body mass index and diabetes duration.
Second, the study could not measure how many of the 81 coded cases caused partial or permanent blindness. Third, the readable abstract omits total enrollment, drug-by-drug counts, and dose breakdowns.
Mounjaro and Zepbound contain tirzepatide, a dual GIP and GLP-1 receptor agonist. The abstract does not isolate tirzepatide or measure weight-loss use without diabetes.
GLP-1 eye side effects: labels, regulators, and urgent symptoms
A June 2025 European Medicines Agency safety review called NAION a "very rare" semaglutide side effect ("up to 1 in 10,000 people"). EMA also noted an epidemiological estimate of about one extra case per 10,000 person-years. The agency stated: "If NAION is confirmed, treatment with semaglutide should be stopped."
As of October 17, 2026, the day this article was written, U.S. eye societies and foreign regulators differ. The American Academy of Ophthalmology and the North American Neuro-Ophthalmology Society issued a September 22, 2026 statement. The two U.S. societies noted that "Causation has not been established."
Instead, U.S. societies favor an individual decision with your doctors if NAION occurs. They oppose routine pre-treatment NAION screening while still backing regular diabetic eye exams. They also rejected Europe's universal stopping rule: "The Academy and NANOS do not support such a blanket recommendation for all patients."
Outside the U.S., a July 23, 2026 Australian Therapeutic Goods Administration safety update added class-wide warnings in Australia, including tirzepatide. Its advisory committee noted evidence "may support" a semaglutide signal, not yet a tirzepatide signal.
Current U.S. labels on DailyMed do not name NAION. Instead, the Ozempic prescribing information (Revised: 5/2026) and Wegovy prescribing information (Revised: 6/2026) warn about diabetic retinopathy, or retinal blood-vessel damage. In Ozempic's two-year diabetes trial, retinopathy complications were 3.0% versus 1.8% on placebo (8.2% vs. 5.2% with prior retinopathy).
The Mounjaro prescribing information and Zepbound prescribing information (both Revised: 8/2026) similarly warn about temporary retinopathy worsening. See our guides to tirzepatide and diabetic retinopathy and the Ozempic vision loss lawsuit.
Even with a small population risk, sudden vision loss is always a medical emergency. A North American Neuro-Ophthalmology Society patient brochure notes that "no proven treatment for NAION" exists. NANOS instructs patients: "Talk with your eye doctor or go to the emergency department immediately if you suddenly lose vision."
Never wait for a routine callback or your next scheduled visit. Australia's TGA similarly directs doctors: "Advise patients to seek urgent medical attention if they experience sudden vision loss, including partial loss of vision."
What to watch next and how prescribed care works
Future studies with specialist-confirmed diagnoses and drug-specific comparisons could make the risks clearer. Any change to U.S. label warnings would also matter. Semaglutide and tirzepatide prescribed through Promise are dispensed by a licensed U.S. compounding pharmacy. Each is dispensed as a compounded medication, which is different from an FDA-approved product: the formulation offered here is not FDA-approved.
Per FDA compounding guidance, the FDA does not verify the safety, effectiveness, or quality of compounded drugs before marketing.
Even so, a licensed provider may still prescribe — that decision is between you and your doctor. At Promise, a licensed provider reviews every request and not everyone qualifies. A prescribed route provides a named clinician, an identified licensed U.S. compounding pharmacy, a labeled preparation and a clear contact for problems. A vial obtained without a prescription leaves out that prescribing review. These safeguards do not remove medicine-related eye risk.
