A nationwide study in Sweden has suggested only a slightly higher risk for non-arteritic anterior ischaemic optic neuropathy (NAION) among users of glucagon-like peptide-1 receptor agonists (GLP-1RAs) compared with sodium–glucose cotransporter-2 (SGLT-2) inhibitor use. However, said the researchers, absolute risks were small, and differences in risk may be partly due to differences in patients’ underlying health.
The study is published in Annals of Internal Medicine.
The study team – from the Karolinska Institutet – used Swedish health data of patients aged 35 to 84 with type 2 diabetes between 2013 and 2024 to emulate a pragmatic target trial assessing the effect of GLP-1RA use on risk for NAION, a condition that causes sudden vision loss in one eye.
Of 107 518 GLP-1RA initiators and 185 898 SGLT-2 inhibitor initiators, the risk for anterior ischaemic optic neuropathy was 0.04% and 0.02% after one year, respectively.
While the relative risk was higher in the GLP-1 group, differences were smaller in analyses that accounted more closely for diabetes severity.
The authors conclude that there may be an association influenced by other patient factors.
Study details
Glucagon-like Peptide-1 Receptor Agonists and Risk for Anterior Ischemic Optic Neuropathy: A Nationwide Cohort Study
Peter Ueda, Henrik Svanström, Jonas Söderling et al.
Published in Annals of Internal Medicine on 14 July 2026
Abstract
Background
There are concerns about a possible link between glucagon-like peptide-1 receptor agonists (GLP-1RAs) and non-arteritic anterior ischaemic optic neuropathy (NAION).
Objective
To examine whether use of GLP-1RAs increases risk for AION, which predominantly comprises NAION.
Design
Nationwide, register-based, cohort study.
Patients
Initiators of GLP-1RAs compared with initiators of sodium–glucose cotransporter-2 (SGLT-2) inhibitors.
Measurements
Anterior ischaemic optic neuropathy in the national patient register. Adjusted risk differences (RDs) and risk ratios (RRs) were estimated using propensity score weighting.
Results
Median follow-up was 1.6 years (IQR, 0.7 to 3.1 years) for GLP-1RA users and 1.5 years (IQR, 0.7 to 2.9 years) for SGLT-2 inhibitor users. Sixty-two of 107 518 GLP-1RA users and 64 of 185 898 SGLT-2 inhibitor users experienced AION. Risks were 0.04% versus 0.02% (RD, 0.02% [95% CI, 0.00% to 0.03%]; RR, 1.93 [CI, 1.00 to 3.73]) at 1 year and 0.12% versus 0.07% (RD, 0.05% [CI, 0.00% to 0.10%]; RR, 1.69 [CI, 0.95 to 3.01]) at 5 years. The differences were substantially attenuated in analyses restricted to patients receiving metformin at baseline (1 year: RD, 0.01% [CI, −0.01% to 0.02%]; RR, 1.40 [CI, 0.64 to 3.05]; 3 years: RD, 0.01% [CI, −0.02% to 0.04%]; RR, 1.24 [CI, 0.68 to 2.26]; 5 years: RD, 0.02% [CI, −0.04% to 0.08%]; RR, 1.23 [CI, 0.65 to 2.33]).
Limitation
Few outcome events, unmeasured confounding, and limited generalisability to GLP-1RA users without diabetes.
Conclusion
The relative risk for AION was higher with GLP-1RA use compared with SGLT-2 inhibitor use in type 2 diabetes. However, absolute risks were small, and the RDs were substantially reduced in analyses restricted to patients receiving metformin to better account for confounding by diabetes severity, suggesting observed increases in risk may reflect residual confounding.
See more from MedicalBrief archives:
Nordisk slams study linking rare blindness to semaglutide
Vision problems flagged after GLP-1 agonists – Utah survey