Study finds no link between semaglutide and a blinding form of macular degeneration

A retrospective study drawing on data from 12 databases found no meaningful association between semaglutide use and neovascular age-related macular degeneration (NVAMD), a finding that pushes back against some earlier research suggesting a possible connection.

The study, conducted through the Observational Health Data Sciences and Informatics (OHDSI) network and published in Ophthalmology, analyzed records from 227,971 new users of semaglutide — the active ingredient in Ozempic and Wegovy — alongside hundreds of thousands of patients on 5 other diabetes medications. The study period ran from December 2017 through December 2024.

“Both the incidence proportion and incidence rate of NVAMD in our study were lower than previously reported,” the study authors, led by Cindy X. Cai, MD, MS, from Wilmer Eye Institute, Johns Hopkins School of Medicine in Baltimore, Maryland; and Biomedical Informatics and Data Science, Division of General Internal Medicine, Department of Medicine, Johns Hopkins University School of Medicine in Baltimore, Maryland, stated. “The estimated annual incidence rate of NVAMD among White adults ≥50 years is 1.8 per 1000.² Because age-related macular degeneration (AMD) affects older adults, and our cohort includes younger patients, a lower overall incidence is expected. However, even when restricting to patients ≥50 years in our cohorts, the incidence rate was still lower than reported. This may reflect the diversity of databases included in our study, for example, the Veterans Affairs (VA), which includes a high proportion of non-Hispanic Black and Hispanic individuals who have a lower AMD incidence³ than non-Hispanic White patients.

Why this was studied

Semaglutide and other GLP-1 receptor agonists have grown rapidly in use for type 2 diabetes and obesity. Prior observational studies produced conflicting results on whether the drug class affects the risk of NVAMD, a serious condition in which abnormal blood vessels grow beneath the retina and can cause significant vision loss. Some studies found increased risk; others found decreased risk. The researchers set out to examine whether those conflicting conclusions were a product of differing data sources, study designs or analytical choices.

How the study was designed

Researchers used 2 methodological approaches. The first was an active-comparator cohort design, comparing new semaglutide users to patients on other diabetes drugs — dulaglutide, exenatide, empagliflozin, sitagliptin and glipizide — using propensity score matching to account for baseline differences between patient groups. The second was a self-controlled case-series analysis, in which patients with NVAMD served as their own controls, comparing their rate of NVAMD diagnosis during periods of semaglutide exposure versus periods without it.

Two definitions of NVAMD were used to test the robustness of findings: one based on diagnosis codes alone, and one requiring both a diagnosis code and a recorded intravitreal anti-VEGF injection within 30 days.

What they found

Across both study designs and both outcome definitions, no statistically significant difference in NVAMD risk was found between semaglutide users and users of any comparator medication. In the self-controlled case-series analysis, the incidence rate ratio for semaglutide was 0.92 (95% CI 0.67 to 1.26) using the diagnosis-only definition and 1.02 (95% CI 0.76 to 1.36) using the diagnosis-plus-procedure definition — neither statistically significant.

One isolated finding in a sensitivity analysis restricted to 2021–2024 showed a lower risk of NVAMD among semaglutide users compared to empagliflozin users using the procedure-inclusive definition. The authors noted, however, that this result would not have survived correction for multiple comparisons given the number of pre-planned analyses in the study.

How this compares to prior research

The authors addressed several conflicting studies directly. One Canadian study had found more than a 2-fold increased risk of NVAMD among GLP-1RA users compared to an unexposed group. The authors wrote that their confidence intervals make such a large harmful effect unlikely, while stopping short of ruling out a small effect in either direction. Other studies using the TriNetX database had found approximately a 20% decreased risk of NVAMD among GLP-1RA users. The authors attributed the divergence across studies to differences in databases, study designs, comparison groups and analytical approaches.

Limitations

The authors noted several constraints. The study relied on diagnosis and procedure codes, which may imperfectly represent clinical reality. NVAMD diagnoses depend on patients actually accessing eye care, which was not directly measured. Although the overall patient population was large, the absolute number of NVAMD cases was relatively small. The study also did not account for cumulative medication doses or examine whether effects differed by age subgroup. Most databases skewed toward populations that may not reflect the full demographic range of semaglutide users.

The authors described the study as "one of the largest" to examine this question and said its findings "provide evidence against large differences in the risk of NVAMD associated with semaglutide use among patients with T2D."

References

1. Cai CX, Toy B, Martin B, et al. Semaglutide and neovascular age-related macular degeneration among adults with type 2 diabetes: An OHDSI network study. Ophthalmology. 2026. doi:10.1016/j.ophtha.2026.05.034

2. Rudnicka AR, Kapetanakis VV, Jarrar Z, et al. Incidence of late-stage age-related macular degeneration in American Whites: Systematic review and meta-analysis. Am J Ophthalmol. 2015;160(1):85-93.e3. doi:10.1016/j.ajo.2015.04.003

3. Rein DB, Wittenborn JS, Burke-Conte Z. Prevalence of age-related macular degeneration in the US in 2019. JAMA Ophthalmol. 2022;140(12):1202-1208. doi:10.1001/jamaophthalmol.2022.4401