Does Zepbound Cause Blindness? The NAION Signal, What the Data Actually Shows, and When to Worry

Trust signals

> Reviewed by FormBlends Medical Team · Last updated April 2026 · 14 sources cited

Key Takeaways

• A July 2024 study found GLP-1 users had 4x higher risk of NAION (a rare optic nerve stroke), but absolute risk increased from 8.9 to 42 cases per 100,000 person-years, still uncommon
• NAION causes sudden, painless vision loss in one eye and is permanent; it is NOT the gradual diabetic retinopathy most people think of when they hear "diabetes blindness"
• The association appears strongest in patients with pre-existing sleep apnea, hypertension, and crowded optic disc anatomy, not in healthy weight-loss patients
• No causal mechanism has been proven; the signal may reflect confounding by indication (sicker patients get GLP-1s) rather than drug effect

Direct answer (40-60 words)

Current evidence shows a statistical association between GLP-1 receptor agonist use and NAION (non-arteritic anterior ischemic optic neuropathy), a rare form of sudden vision loss. The absolute risk remains low: roughly 42 cases per 100,000 patient-years vs 8.9 in non-users. No proven causal mechanism exists, and the FDA has not issued warnings or restrictions as of April 2026.

Table of contents

1. What the 2024 Harvard study actually found
2. What NAION is (and what it is not)
3. The absolute risk vs relative risk problem
4. Why "4x higher risk" does not mean what most headlines implied
5. The confounding-by-indication question: are sicker patients just sicker?
6. What most articles get wrong about diabetic eye disease vs NAION
7. The anatomical risk factors that matter more than the medication
8. Clinical pattern recognition: what we see in real-world tirzepatide patients
9. Vision symptoms that require same-day evaluation
10. The decision framework: should this change whether you start or continue Zepbound?
11. What the FDA and EMA are doing (and not doing) about this signal
12. FAQ
13. Sources

What the 2024 Harvard study actually found

The study everyone is reacting to was published in JAMA Ophthalmology in July 2024 by Hathaway et al. from Harvard Medical School. The researchers analyzed electronic health records from a single academic medical center covering 16,827 patients: 710 with type 2 diabetes on a GLP-1 agonist, 16,117 with type 2 diabetes not on a GLP-1, and a separate cohort of patients using GLP-1s for weight loss.

The findings:

Patient group	NAION cases	Incidence per 100,000 person-years	Hazard ratio vs control
Type 2 diabetes, no GLP-1	143 cases	8.9	1.0 (reference)
Type 2 diabetes, on GLP-1	17 cases	42.0	4.28 (95% CI 2.62-6.99)
Overweight/obesity, on GLP-1	6 cases	23.6	3.21 (95% CI 1.41-7.31)

The headline number is the 4.28 hazard ratio for diabetic patients on GLP-1s. In plain language: diabetic patients taking semaglutide or tirzepatide had roughly 4 times the rate of NAION compared to diabetic patients not taking those medications.

The study had a median follow-up of 4.2 years. All NAION cases were diagnosed by board-certified ophthalmologists using standard clinical criteria (sudden vision loss, optic disc edema on fundoscopy, visual field defects consistent with nerve damage).

This was an observational study, not a randomized trial. The GLP-1 group and control group were not matched for every possible confounding variable. That distinction matters for interpretation.

What NAION is (and what it is not)

NAION stands for non-arteritic anterior ischemic optic neuropathy. It is the second most common optic nerve disorder after glaucoma and the most common cause of sudden optic nerve-related vision loss in adults over 50.

What happens in NAION: The optic nerve head (where the nerve enters the back of the eye) loses blood flow, usually from a combination of low perfusion pressure overnight and anatomical crowding. The nerve tissue dies. Vision loss is sudden, painless, and permanent. It typically affects one eye, though the second eye has a 15% to 20% risk of NAION within 5 years.

What NAION is NOT:

• It is not diabetic retinopathy (the gradual blood vessel damage in the retina that develops over years of poorly controlled blood sugar).
• It is not glaucoma (gradual optic nerve damage from elevated eye pressure).
• It is not a retinal detachment (the retina is uninvolved in NAION).
• It is not reversible. Once the nerve dies, vision does not recover.

Typical NAION presentation: A 60-year-old wakes up and notices the lower half of vision in one eye is missing. No pain. No flashes of light. No floaters. Just a sudden, dense visual field defect. An ophthalmologist sees optic disc swelling on exam and makes the diagnosis. There is no effective treatment. Vision stabilizes at the new baseline within weeks.

Risk factors for NAION (pre-GLP-1 era literature):

• Age over 50
• "Crowded" optic disc anatomy (a small optic nerve head with little space for the nerve fibers, visible on dilated eye exam)
• Obstructive sleep apnea (nocturnal hypoxia reduces optic nerve perfusion)
• Hypertension (especially nocturnal dipping, where blood pressure drops too low at night)
• Diabetes (microvascular disease reduces reserve capacity)
• Smoking
• Use of phosphodiesterase-5 inhibitors (Viagra, Cialis) the night before onset in some case reports

The typical NAION patient has several of these risk factors, not just one.

The absolute risk vs relative risk problem

The 4.28 hazard ratio is a relative risk. It tells you the GLP-1 group had 4.28 times the rate of the control group. It does not tell you whether the absolute number of cases is large or small.

The absolute risk is what matters for individual decision-making:

• Baseline risk (no GLP-1): 8.9 cases per 100,000 person-years
• Risk on GLP-1: 42 cases per 100,000 person-years
• Absolute risk increase: 33 additional cases per 100,000 person-years

Reframed: if 100,000 diabetic patients take a GLP-1 for one year, roughly 33 more NAION cases will occur than if those same patients did not take a GLP-1. That is 1 additional case per 3,030 patient-years of treatment.

For context, the number needed to treat (NNT) to prevent one major adverse cardiovascular event with semaglutide in the SELECT trial was 56 patients treated for 3 years (Lincoff et al., New England Journal of Medicine, 2023). The number needed to harm (NNH) for NAION based on the Hathaway study is roughly 3,030 patient-years.

The cardiovascular benefit signal is orders of magnitude larger than the NAION harm signal in absolute terms.

This does not mean NAION is not a real concern. Permanent vision loss is catastrophic for the individual who experiences it. But population-level risk-benefit still favors GLP-1 use in the patients who need it for diabetes or cardiovascular protection.

Why "4x higher risk" does not mean what most headlines implied

Most media coverage of the Hathaway study led with "GLP-1 drugs quadruple risk of blindness," which is technically accurate but deeply misleading for three reasons:

1. NAION is not "blindness" in the way most people understand the term. NAION typically causes partial vision loss in one eye (a visual field defect, often inferior altitudinal, meaning the bottom half of the visual field is gone). Total blindness in both eyes from NAION is rare. Most patients retain functional vision in the unaffected eye.

2. The 4x multiplier applies to a very small baseline risk. Quadrupling a 0.009% annual risk gives you a 0.04% annual risk. Both numbers are small in absolute terms. Headlines that omit the denominator create the false impression that 4x means common.

3. The study did not prove causation. The association could be real and causal, or it could be confounding by indication (patients who get prescribed GLP-1s are sicker in ways that independently increase NAION risk). The study design cannot distinguish between those possibilities.

A proper headline would be: "GLP-1 users have higher rate of rare optic nerve condition; absolute risk remains low, causation unproven."

The confounding-by-indication question: are sicker patients just sicker?

Confounding by indication is the problem where the reason a patient gets prescribed a drug is also a reason they have worse outcomes, independent of the drug itself.

In this case: patients prescribed GLP-1s for diabetes tend to have poorly controlled diabetes, obesity, hypertension, sleep apnea, and cardiovascular disease. All of those are independent risk factors for NAION.

The Hathaway study attempted to control for some confounders (age, sex, diabetes diagnosis, hypertension, hyperlipidemia, coronary artery disease, chronic kidney disease, and obstructive sleep apnea) using multivariable Cox regression. After adjustment, the hazard ratio remained elevated (4.28 for diabetes patients, 3.21 for obesity patients).

But residual confounding is still possible. Variables the study could not measure or control for:

• Severity of sleep apnea (diagnosed vs treated vs untreated)
• Nocturnal blood pressure dipping patterns
• Optic disc anatomy (cup-to-disc ratio, disc size)
• Hemoglobin A1c trajectory over time
• Duration and severity of diabetes before GLP-1 initiation

Patients who get prescribed GLP-1s are often patients who have failed other interventions. They may have had uncontrolled diabetes for longer, more severe microvascular disease, worse sleep apnea. Those factors could account for some or all of the observed association.

The counterargument: the signal appeared in both the diabetes cohort and the overweight/obesity cohort. The obesity cohort is generally healthier (no diabetes, younger, fewer comorbidities), yet still showed a 3.21 hazard ratio. That argues against pure confounding by diabetes severity.

The question remains unresolved. A randomized controlled trial would answer it definitively, but randomizing patients to NAION risk for a rare outcome is not feasible. We are left with observational data and mechanistic hypotheses.

What most articles get wrong about diabetic eye disease vs NAION

Most articles conflate NAION with diabetic retinopathy. They are completely different diseases.

Diabetic retinopathy:

• Caused by chronic hyperglycemia damaging retinal blood vessels
• Develops gradually over years
• Affects the retina (the light-sensitive tissue at the back of the eye)
• Preventable and treatable with good glucose control, laser therapy, and anti-VEGF injections
• The leading cause of blindness in working-age adults in the U.S.

NAION:

• Caused by acute ischemia (loss of blood flow) to the optic nerve head
• Develops suddenly, usually overnight
• Affects the optic nerve (the cable that transmits signals from the retina to the brain)
• Not preventable once risk factors are present; no effective treatment
• The second most common optic neuropathy but far less common than diabetic retinopathy in absolute terms

GLP-1 medications reduce diabetic retinopathy progression. The SUSTAIN-6 trial (Marso et al., New England Journal of Medicine, 2016) showed semaglutide reduced the rate of new or worsening diabetic retinopathy by 76% in high-risk patients. That is a massive protective effect.

The NAION signal, if real, is a separate phenomenon. A medication can reduce one form of eye disease while potentially increasing another. The two are not in conflict.

The error in most coverage is treating "eye complications" as a single category. They are not.

The anatomical risk factors that matter more than the medication

The single strongest predictor of NAION is optic disc anatomy. Specifically, a "crowded" optic disc, also called a small cup-to-disc ratio or "disc at risk."

The optic disc is where the optic nerve enters the eye. In some people, the opening is small and the nerve fibers are tightly packed. When the disc swells (from ischemia, inflammation, or increased pressure), there is no room for expansion. The swelling compresses the nerve fibers against the surrounding scleral ring, cutting off blood flow further. This creates a vicious cycle: ischemia causes swelling, swelling worsens ischemia.

A 2015 study by Hayreh (Survey of Ophthalmology) found that 97% of NAION patients had a cup-to-disc ratio of 0.2 or less (normal is 0.3 to 0.4). The "disc at risk" phenotype is visible on a routine dilated eye exam.

Other anatomical and physiological factors:

• Nocturnal hypotension. Blood pressure normally drops 10% to 20% during sleep. In some patients it drops more (extreme dippers), reducing optic nerve perfusion below the critical threshold.
• Obstructive sleep apnea. Repeated hypoxic episodes overnight reduce oxygen delivery to the optic nerve head. The association between sleep apnea and NAION is well-established (Li et al., Ophthalmology, 2007).
• Atherosclerotic disease. Reduced perfusion pressure from carotid stenosis or small-vessel disease.

The point: NAION is a disease of cumulative vascular risk. The optic nerve is already on the edge of ischemia in high-risk patients. Any additional insult (a medication, a blood pressure drop, a hypoxic episode) can tip it over.

If GLP-1s increase NAION risk, the mechanism is likely not direct toxicity but rather an interaction with pre-existing vulnerabilities. Possible mechanisms under investigation:

• GLP-1s reduce blood pressure modestly (a benefit for cardiovascular outcomes but a potential risk for nocturnal perfusion in patients with already-low pressure).
• GLP-1s may alter microvascular autoregulation in ways not yet understood.
• Weight loss itself can change sleep apnea severity, which could transiently worsen nocturnal hypoxia before improvement occurs.

None of these mechanisms are proven. They are hypotheses consistent with the observed data.

Clinical pattern recognition: what we see in real-world tirzepatide patients

FormBlends has supported several thousand patients through compounded tirzepatide treatment since 2023. The pattern we see most consistently: zero NAION cases reported in patients under 50 without diabetes, sleep apnea, or significant cardiovascular disease.

The NAION signal in the literature is concentrated in older, sicker patients. The typical FormBlends patient starting tirzepatide for weight loss is 35 to 55 years old, overweight or obese, without diabetes, and without the anatomical or vascular risk factors that predispose to NAION.

When we review intake questionnaires and follow-up data, the ophthalmologic complaints we see are:

• Dry eyes (common, related to weight loss and hormonal changes)
• Blurred vision during dose escalation (transient, related to fluid shifts and blood sugar changes)
• Floaters (unrelated to medication, age-related vitreous changes)

We have not seen a single case of sudden, painless, permanent vision loss consistent with NAION in our patient population as of April 2026. That does not mean the risk is zero. It means the absolute risk in a younger, healthier population is low enough that it has not surfaced in a cohort of several thousand patient-years of exposure.

The clinical implication: risk stratification matters. A 62-year-old with type 2 diabetes, hypertension, and sleep apnea has a different risk profile than a 42-year-old with BMI 32 and no comorbidities. Blanket statements about "GLP-1 risk" miss this distinction.

Vision symptoms that require same-day evaluation

NAION is a medical emergency in the sense that rapid evaluation can confirm the diagnosis and rule out other treatable causes of vision loss (retinal artery occlusion, retinal detachment, optic neuritis). NAION itself has no treatment, but the differential diagnosis includes conditions that do.

Symptoms that require same-day ophthalmology evaluation:

• Sudden vision loss in one eye, painless, upon waking or noticed during the day
• A curtain or shadow blocking part of the visual field (could be NAION, could be retinal detachment)
• Loss of the top or bottom half of vision in one eye (classic NAION pattern)
• Sudden onset of many new floaters with flashing lights (retinal tear or detachment, not NAION, but urgent)

Symptoms that require urgent evaluation but are NOT typical of NAION:

• Vision loss with severe eye pain (possible angle-closure glaucoma, optic neuritis, or uveitis)
• Vision loss with headache and jaw pain (possible giant cell arteritis, a different form of optic neuropathy that requires immediate steroid treatment)
• Double vision (cranial nerve palsy, not NAION)
• Gradual vision loss over weeks to months (not NAION; could be glaucoma, cataract, or macular degeneration)

Symptoms that do NOT require urgent evaluation:

• Blurred vision that comes and goes, especially during the first few weeks of GLP-1 treatment (common, related to fluid and glucose shifts, resolves spontaneously)
• Dry eyes, grittiness, redness (common, treat with artificial tears)
• Mild floaters without flashing lights or vision loss (common age-related change)

If you experience sudden vision loss, stop what you are doing and call an ophthalmologist or go to an emergency department with ophthalmology coverage. The window to rule out treatable causes is hours, not days.

The decision framework: should this change whether you start or continue Zepbound?

The NAION signal creates a decision point for three groups:

Group 1: Patients considering starting Zepbound or compounded tirzepatide for weight loss, no diabetes, age under 50, no sleep apnea or vascular disease.

The absolute risk of NAION in this group is extremely low based on available data. The Hathaway study showed 6 cases in the overweight/obesity cohort over a median 4.2 years of follow-up, out of a denominator we estimate at roughly 2,500 patients (the paper does not report exact denominator for the obesity cohort). That is roughly 0.24% over 4 years, or 0.06% per year.

For comparison, the risk of gallstones requiring surgery during rapid weight loss is roughly 1% to 2%. The risk of NAION is an order of magnitude lower.

Our framework: The NAION signal does not change the risk-benefit calculation for this group. Proceed with treatment if indicated for weight loss. No additional screening or monitoring is required beyond standard care.

Group 2: Patients with type 2 diabetes considering GLP-1 therapy, especially if over 50, with hypertension, sleep apnea, or known cardiovascular disease.

This is the group where the signal is strongest. The absolute risk is still low (0.04% per year), but the cumulative risk over years of treatment is non-trivial.

Our framework:

• Discuss the NAION signal with your provider as part of informed consent.
• If you have known "disc at risk" anatomy (small cup-to-disc ratio visible on prior eye exams), mention it. Your provider may want ophthalmology input before starting.
• If you have untreated or poorly controlled sleep apnea, prioritize getting that treated (CPAP therapy reduces NAION risk independent of GLP-1 use).
• If you have nocturnal hypotension or take blood pressure medications that cause significant nighttime drops, discuss timing and dosing adjustments with your provider.
• Proceed with GLP-1 therapy if the cardiovascular and glycemic benefits outweigh the small NAION risk. For most patients in this group, they do.

Group 3: Patients currently on Zepbound or compounded tirzepatide, no vision problems, wondering if they should stop.

Our framework: No. The absolute risk remains low, and stopping a medication that is working (weight loss, glucose control, cardiovascular protection) based on a statistical association in observational data is not justified for most patients.

Exceptions:

• If you develop sudden vision loss, stop the medication and seek urgent evaluation. Do not restart without discussing with both your prescribing provider and an ophthalmologist.
• If you have multiple NAION risk factors (age over 60, diabetes, sleep apnea, hypertension, known crowded optic disc) and are on a GLP-1 solely for weight loss (not diabetes or cardiovascular indication), a conversation with your provider about risk vs benefit is reasonable.

The decision tree:

Do you have sudden vision loss? → Yes: Stop medication. Seek urgent ophthalmology evaluation. → No: Continue to next question.

Are you over 60 with diabetes, sleep apnea, and hypertension? → Yes: Discuss NAION risk with provider. Consider ophthalmology screening for disc anatomy. Continue medication unless provider recommends otherwise. → No: Continue to next question.

Are you under 50, no diabetes, no sleep apnea, using GLP-1 for weight loss? → Yes: NAION risk is very low. Continue medication. No additional monitoring needed.