Scientific deep-dive

GLP-1 Drugs and Eye Problems: What the Trial Meta-Analyses Show

Pooled randomized trials show no overall rise in eye disorders or diabetic retinopathy with GLP-1 drugs, while the rare optic nerve condition NAION remains an imprecise and contested signal. Here is how the meta-analyses compare and where they fall short.

By Eli Marsden · Founding Editor
Editorially reviewed & fact-checked against primary sources · How we verify contentLast reviewed
8 min read·14 citations

Pooled randomized trials have not shown that GLP-1 drugs raise eye problems overall. The broadest analysis, 78 semaglutide trials with 73,640 participants, put the odds ratio for any eye disorder at 1.01 (95% CI 0.91 to 1.12)[1]. The exception to watch is NAION, a rare infarction of the optic nerve head: the same analysis found an odds ratio of 3.92 (95% CI 1.02 to 15.02), equal to about 1 extra case per 10,000 participants, with low certainty[1]. Most of these trials were not built to look for eye disease, and that limit shapes how far either result can be pushed.

About this article

This page compares the pooled analyses across eye conditions. For one condition in depth, read our pages on NAION and blurred vision during treatment.

The pooled analyses side by side

Four things vary between these papers: the drugs, the populations, the study designs and the outcome definitions. The table keeps them apart. Results come from Natividade 2025[1], Li 2026[2], Nogueira 2026[3], Ortiz-Seller 2026[4] and Anatriello 2026[5]. Comparators are placebo or active drugs in the trials, and other diabetes drugs in Anatriello.

Pooled estimates for GLP-1 drugs and eye outcomes. Values in parentheses are 95% confidence intervals.
AnalysisEvidence baseOutcomeResult
Natividade 202578 RCTs, 73,640 people, semaglutide onlyAny eye disorderOR 1.01 (0.91 to 1.12)
Natividade 2025Same review, trials in people with diabetesDiabetic retinopathyOR 1.04 (0.92 to 1.17)
Natividade 20255 trials reporting NAION, 25,330 peopleNAIONOR 3.92 (1.02 to 15.02)
Li 202620 RCTs, 83,288 people, 76.4% with type 2 diabetes, GLP-1 classOptic nerve or vision-threatening serious eventsOR 1.20 (0.73 to 1.97)
Li 2026SameIschemic optic neuropathyOR 1.50 (0.49 to 4.63)
Nogueira 202615 studies including 8 trials, over 1.5 million people, GLP-1 classAny ocular eventOR 0.95 (0.86 to 1.05)
Nogueira 2026SameNAIONOR 1.70 (1.23 to 2.36)
Ortiz-Seller 202687 RCTs, 125,418 people with type 2 diabetes, 9 drugsDiabetic retinopathyNo statistically credible change for any drug vs placebo
Anatriello 202628 observational studies in type 2 diabetesNAIONRR 1.01 (0.62 to 1.64)
Anatriello 2026SameGlaucomaHR 0.84 (0.71 to 1.00)
Anatriello 2026SameNew retinopathy; progressionHR 0.96 (0.85 to 1.08); HR 0.97 (0.83 to 1.14)

What a neutral result on “any eye disorder” can and cannot say

In absolute terms, Natividade estimated 1 more eye disorder per 1,000 participants on semaglutide (95% CI 6 fewer to 8 more), with moderate certainty[1]. The outcome was broad: any event classified under the MedDRA eye disorders category or reported by investigators as affecting the visual system[1]. A composite that wide tells you the overall burden of reported eye events looks similar. It can't rule out a rare condition with its own pattern, because a handful of cases disappears into thousands of unrelated complaints.

The way trials measured eyes matters too. Nearly half of the trials in Natividade did not perform eye examinations or retinal imaging, only one used optical coherence tomography, and none used campimetry, the visual field test[1]. The abstract rates overall risk of bias as low, but the full paper flags 28 trials as high risk for outcome measurement because no eye exam took place at the study’s end[1]. Li reaches the same limit from a different angle: vision-threatening events were collected as adverse events, not prespecified outcomes[2].

Diabetic retinopathy: strong on average, thin for high-risk patients

Retinopathy has the firmest numbers. Natividade calculated that 20,627 participants were needed to exclude a 20% increase in risk and had 49,368, with an estimated 1 more event per 1,000 (95% CI 6 fewer to 12 more)[1]. Ortiz-Seller, working with 6,330 retinopathy events across nine drugs, found no dose-response for retinopathy and no credible differences between drugs, and its trial sequential analysis excluded a class-wide effect of plus or minus 20%[4].

That does not cancel the older warning. In the 2-year trial behind the SUSTAIN-6 retinopathy signal, complications occurred in 3% on semaglutide and 1.8% on placebo, as the Wegovy label reports[6]. Ortiz-Seller says smaller, drug-specific or high-risk subgroup effects can’t be excluded, and its trial-level exploratory analyses suggested effect modification by baseline retinopathy risk and body mass index[4]. Natividade adds two more gaps: no trial reported retinopathy severity, and people with proliferative retinopathy or maculopathy needing acute treatment were excluded[1]. Those are the patients with the most to lose, and the pooled trials say the least about them.

Patients with a history of diabetic retinopathy should be monitored.
— Wegovy prescribing information, Highlights, Section 5.8 (DailyMed version published June 2026)

NAION: why the estimates refuse to line up

NAION is where the papers disagree, and the scarcity of cases is part of the reason. In Natividade’s five trials reporting it, the estimated incidence was 60.7 per 100,000 on semaglutide and 8.2 per 100,000 in controls[1]. The authors’ trial sequential analysis needed 37,630 people to exclude a number needed to harm of 1,000 and had 25,330, so they call the evidence insufficient for a definitive conclusion[1]. A post hoc version that added trials with zero events in both arms did cross the futility boundary, which shows how much the answer depends on analytic choices[1].

Liu 2026 split the data by indication and design. In randomized trials for diabetes there were 5 ischemic optic neuropathy events on semaglutide against 1 without it (RR 1.76, 95% CI 0.43 to 7.25); in weight-loss trials, 4 against 1 (RR 2.18, 95% CI 0.33 to 14.34)[7]. Both point estimates sit above 1, and neither interval excludes no effect. Its observational data showed HR 1.85 (95% CI 1.20 to 2.85) in diabetes and HR 1.57 (95% CI 0.69 to 3.59) for weight loss[7].

Other syntheses of largely observational data land in different places. Dhivagaran 2026 reported RR 2.52 (95% credible interval 1.56 to 4.72) for semaglutide, with low-to-moderate certainty[8]. Abdelaal 2026 found a pooled adjusted HR of 1.63 (95% CI 0.88 to 2.39), low certainty, with 1.70 (95% CI 1.10 to 2.64) in type 2 diabetes and 0.47 (95% CI 0.19 to 1.13) in obesity or overweight, and warned that the difference by indication may reflect disease rather than drug[9]. Anatriello’s RR of 1.01 carries an I² of 89%, meaning the underlying studies varied widely[5]. These papers used different comparators and search dates, so they should not be read as one number.

Absolute risk stays small in every version. Nogueira puts it at 0.09% in GLP-1 users, a risk difference of 0.037%, or a number needed to harm of about 2,700[3]. None of this establishes causation. Abdelaal says so directly and notes that NAION’s major risk factors overlap with the populations prescribed semaglutide[9]. European regulators have recommended listing NAION as a very rare adverse event for semaglutide[5].

Evidence on other conditions is almost all retrospective, and what looks like protection deserves the same skepticism as what looks like harm. For glaucoma, Anatriello’s HR of 0.84 has an upper limit of 1.00[5]. Amaral 2025 found a significant reduction only after removing one study (HR 0.71, 95% CI 0.60 to 0.85), and its authors asked for cautious interpretation[10].

For diabetic macular edema, 13 retrospective cohorts showed no increase against mixed diabetes drugs (HR 0.81, 95% CI 0.52 to 1.26), but a higher relative risk against SGLT2 inhibitors (HR 1.50, 95% CI 1.17 to 1.94), with very low certainty overall[11]. A TriNetX cohort of 68,536 nondiabetic adults aged 60 or older with overweight or obesity reported much lower 5-year risks with liraglutide or semaglutide than with other weight-loss drugs, including cataract (RR 0.45, 95% CI 0.40 to 0.50) and macular degeneration (RR 0.33, 95% CI 0.23 to 0.48)[12]. Effects that large across so many unrelated conditions in database research are a reason to suspect residual confounding before crediting the drug. This was an association from a matched retrospective cohort, not a trial.

Is this a semaglutide question or a class question?

Unresolved. A 2026 narrative review argues that receptor signaling differences could explain why longer-acting semaglutide has drawn more concern than liraglutide or exenatide, but it offers a hypothesis with testable predictions, not patient outcome data[13]. On retinopathy, the network analysis found no credible between-drug differences[4]. Direct evidence on tirzepatide is thin: a retrospective cohort of 121 records found combined retinal events in 2 of 14 semaglutide users, 10 of 16 tirzepatide users and 46 of 91 unexposed patients, and its author says these exploratory figures do not establish comparative safety[14].

If you take a GLP-1 for weight loss, not diabetes

Most of this evidence comes from diabetes. Anatriello’s studies all involved type 2 diabetes, and the authors could not run a separate overweight or obesity analysis[5]. Li’s trial population was 76.4% diabetic[2]. Natividade included 13 trials of obesity or overweight without diabetes, and 2 of its 5 NAION trials came from that group, with the odds ratio higher in the diabetes trials[1]. The retinopathy findings mainly concern people with diabetes. Liu’s weight-loss trials had 5 ischemic optic neuropathy events in total[7], so the NAION question has few events behind it for anyone taking these drugs for weight alone.

What to raise with a prescriber

  • Any history of diabetic retinopathy. Ask whether an eye exam should come before starting or raising a dose. The Wegovy label calls for monitoring these patients[6], and both Ortiz-Seller and Anatriello support continued surveillance[4][5].
  • NAION risk factors or a prior episode. Hypertension, diabetes, hyperlipidemia and smoking are the major risk factors Natividade lists[1]. Trials offer little guidance for someone with a previous NAION, so that deserves a specific conversation.
  • Sudden loss of vision, even without pain. NAION typically presents as abrupt, painless vision loss[1], and it is a serious cause of sudden, often permanent vision loss[7]. Treat it as an emergency and contact an eye doctor the same day, then tell your prescriber.

Frequently Asked Questions

Across randomized trials, overall eye events look about the same with and without these drugs. A meta-analysis of 78 semaglutide trials found an odds ratio of 1.01 for any eye disorder. The exceptions are a possible rise in a rare optic nerve condition called NAION and early worsening of existing diabetic retinopathy in some people with type 2 diabetes.
It appears rare. One meta-analysis of randomized trials estimated about 1 extra NAION case per 10,000 participants, with low certainty. Another analysis across trials and cohorts of GLP-1 drugs reported an absolute risk of 0.09% and a number needed to harm of about 2,700. Causation has not been established.
On average, pooled trials found no overall increase (odds ratio 1.04), and that analysis had enough participants to exclude a 20% rise. But the SUSTAIN-6 trial found complications in 3% on semaglutide vs 1.8% on placebo, the trials excluded people with the most advanced disease, and high-risk subgroup effects can't be ruled out. Anyone with existing retinopathy should raise monitoring with their prescriber.
That depends on your history, so ask your prescriber. The clearest case is a person with type 2 diabetes or known retinopathy, because the Wegovy label says patients with a history of diabetic retinopathy should be monitored. Researchers also note that most trials were not designed to assess eye events, so the trial data offer limited guidance on routine screening.

References

  1. 1.Natividade GR, Spiazzi BF, Baumgarten MW, Bassotto C, Pereira AA, Fraga BL, Scalco BG, Mattes NR, Lavinsky D, Kramer CK, Gerchman F. Ocular Adverse Events With Semaglutide: A Systematic Review and Meta-Analysis. JAMA Ophthalmol. 2025. PMID: 40810985.
  2. 2.Li HY, Chan TK, Co Shih K, Cheung BMY, Yiu KH, Tse HF, Chan YH. GLP-1 Receptor Agonists and Risk of Optic Nerve or Vision-Threatening Events in Patients With Type 2 Diabetes or Cardiometabolic Diseases: A Meta-analysis of Randomized Controlled Trials. Diabetes Care. 2026. PMID: 41587563.
  3. 3.Nogueira A, Rassi TNO, Iqbal A, Felix N, Alghaith O, Khan A, Rassi N, Maia M, Moura FA. Association Between GLP-1 Receptor Agonists and Ischemic Optic Neuropathy: A Meta-analysis. Diabetes Care. 2026. PMID: 41498755.
  4. 4.Ortiz-Seller A, Real JT, Morcillo E, Ortiz JL. Glucagon-like peptide-1 receptor agonists and risk of diabetic retinopathy in type 2 diabetes: A systematic review and network meta-analysis of randomized clinical trials. Surv Ophthalmol. 2026. PMID: 42612754.
  5. 5.Anatriello A, Liguori V, Cagnotta C, Pentella C, Scavone C, Capuano A, Rinaldi B. Ocular disorders during treatment with GLP-1 receptor agonists: a systematic review and meta-analysis of observational studies. Front Pharmacol. 2026. PMID: 42311413.
  6. 6.Novo Nordisk. WEGOVY (semaglutide) injection and tablet. Prescribing information. DailyMed, National Library of Medicine. 2026. https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=ee06186f-2aa3-4990-a760-757579d8f77b
  7. 7.Liu Z, Raveendran D, Fourlanos S, Chong EW. Rate and Risk of Nonarteritic Anterior Ischemic Optic Neuropathy with Semaglutide Use for Diabetes and Weight Loss: A Systematic Review and Meta-Analysis. Ophthalmology. 2026. PMID: 41475544.
  8. 8.Dhivagaran T, Butt F, Arunasalam L, Bae I, Mikhail D, Tao BK, Xie JS, Balas M, Popovic M, Margolin E. Glucagon-like Peptide-1 Receptor Agonists and Risk of Nonarteritic Anterior Ischemic Optic Neuropathy: Systematic Review and Meta-Analysis. Neurology. 2026. PMID: 42060879.
  9. 9.Abdelaal A, Abu Serhan H, Alsaadi M, Yaldo L, Gaier ED, Elhusseiny AM. Semaglutide and the Risk of Nonarteritic Ischemic Optic Neuropathy: A Systematic Review and Certainty of Evidence Meta-Analysis. Ophthalmology. 2026. PMID: 41692115.
  10. 10.Amaral DC, Guedes J, Cruz MRB, Cheidde L, Nepomuceno M, Magalhães PLM, Brazuna R, Mora-Paez DJ, Huang P, Razeghinejad R, Schuman JS, Myers JS. GLP-1 Receptor Agonists Use and Incidence of Glaucoma: A Systematic Review and Meta-Analysis. Am J Ophthalmol. 2025. PMID: 39732312.
  11. 11.Al-Burak SA, Voznyy V, Elganga M, Zajner C, Juncal V, Sheidow TG, Malvankar-Mehta MS. Diabetic macular edema and GLP-1 receptor agonist use: a systematic review and meta-analysis. Can J Ophthalmol. 2026. PMID: 42242293.
  12. 12.Chen KE, Wang PC, Lin HA, Lin SF. Risk of Age-Related Ocular Diseases in Non-Diabetic Adults With Obesity Using Glucagon-Like Peptide 1 Receptor Agonists. Diabetes Obes Metab. 2026. PMID: 42259622.
  13. 13.Li H, Peng S, Lei D. Divergent ocular safety profiles of glucagon-like peptide-1 receptor agonists: a pharmacological heterogeneity framework. Front Med (Lausanne). 2026. PMID: 42866022.
  14. 14.Muhsen J. Diabetic Retinal Disease Onset and Progression During Established Semaglutide or Tirzepatide Treatment: A Retrospective Comparative Cohort Study With a Descriptive Regional Nonarteritic Anterior Ischemic Optic Neuropathy (NAION) Review. Cureus. 2026. PMID: 42909983.

GLP-1, Ozempic & Vision Loss: The NAION Evidence

Does Ozempic cause blindness? Studies link semaglutide to NAION (sudden one-eye vision loss) — relative risk is real (HR ~1.8–8) but absolute risk is low (≤1 in 10,000 per EMA). Causality unproven; diabetes itself harms the optic nerve. Know the one red-flag symptom.

10 min read

Semaglutide vs Tirzepatide: Which Has Higher NAION Risk?

Data lean toward a smaller NAION signal with tirzepatide than semaglutide, but the evidence is observational and no trial compares them. What cohorts and adverse event reports show, why report databases can't measure the gap, and what to ask a prescriber.

8 min read

GLP-1 and Glaucoma: NAION Signal, IOP, and Dry Eye Evidence

The Hathaway 2024 JAMA Ophthalmology study flagged a NAION signal with semaglutide. We review the actual data, the glaucoma + IOP literature, the dry-eye pattern during dose escalation, and the ophthalmology follow-up protocol.

11 min read

GLP-1 Exposure Around Conception: Birth Defect Meta-Analyses

Pooled observational data show no clear rise in major birth defects after GLP-1 exposure around conception, but certainty is very low. See what the 2026 meta-analyses found, and why one large cohort shapes the results.

9 min read

Do GLP-1 Drugs Lower Obesity-Related Cancer Risk? What Meta-Analyses Show

Observational reviews link GLP-1 drugs to lower rates of several obesity-related cancers, but a meta-analysis of 48 randomized trials found little or no clear effect. Why the evidence splits, and why the comparison drug changes the answer.

8 min read

Do Weight-Loss Drugs Improve Natural Fertility? What the Meta-Analyses Show

GLP-1 therapy was linked to more natural pregnancies than metformin in PCOS trials, but pooled data are small, overall pregnancy did not differ significantly, and live birth benefit remains unproven. Here is what the reviews, orlistat trials and real-world records show.

7 min read

Where to get semaglutide (Ozempic / Wegovy) safely: vetted online providers

Vetted telehealth providers that prescribe online. We compare pricing, form, and states served.

No insurance needed · vetted by our editors

WeightLossRankings.org is reader-supported. When you buy through links on our site, we may earn an affiliate commission. How we make money

8.3

Found

Medication-included compounded GLP-1 on 28-day cash plans, plus insurance help for brand-name GLP-1s

Read full review →

From $169/mo

12-month plan rate

Get started →
8.1

Oak

Compounded semaglutide and tirzepatide at a low starting price, with a full price list before you sign up

Read full review →

From $200/mo

starting dose

Get started →
7.8

SnagRx

Compounded semaglutide and tirzepatide at a flat lifetime price

Read full review →

From $99/mo

Get started →

About this analysis

Weight Loss Rankings is an independent editorial publication. Our provider dataset is verified directly against each provider's website and updated continuously. Read our full methodology →