Abstract
This cohort study evaluates the 3-year risk of nonarteritic anterior ischemic optic neuropathy among veterans with type 2 diabetes who initiated glucagon-like peptide-1 receptor agonists (GLP-1RAs) or a sodium-glucose cotransporter-2 (SGLT2) inhibitor.
Introduction
Glucagon-like peptide-1 receptor agonists (GLP-1RA) are increasingly being used for various indications.1,2 Pharmacovigilance reports suggest a disproportionate signal of ocular adverse events and visual impairment among GLP-1RA users, and emerging population-based studies raise concern about an increased risk of nonarteritic anterior ischemic optic neuropathy (NAION).3,4,5 However, findings remain inconsistent,5,6 and it remains unclear whether GLP-1RA use is specifically associated with NAION rather than other optic disorders.
Methods
For this cohort study, we leveraged the nationwide electronic health records of the US Department of Veterans Affairs (VA) to emulate a target trial to compare the 3-year risk of NAION among veterans with type 2 diabetes who initiated GLP-1RAs or sodium-glucose cotransporter-2 inhibitors (SGLT2i) between January 1, 2017, and December 31, 2024 (eTable in Supplement 1). The date of analysis was November 28, 2025.
NAION diagnosis was defined by International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code H47.01*. We validated all ICD-10–identified cases using clinical text documentation, yielding a positive predictive value of 84.70% (95% CI, 83.10–86.20). We evaluated alternative definitions including NAION diagnosis by an eye care specialist, repeated NAION diagnoses, and specialist diagnosed NAION with repeated diagnosis (eMethods in Supplement 1).
We balanced the GLP-1RA and SGLT2i groups through propensity score–based inverse probability weighting to estimate the intention-to-treat average treatment effect within the GLP-1RA group. Weighted Fine and Gray subdistribution hazard models, accounting for death as a competing risk, were used to estimate the 3-year cumulative incidence, cumulative incidence difference (CID), and cumulative incidence ratio (CIR). A 95% CI for the CIR excluding 1 was considered statistically significant.
To evaluate the specificity of the association between GLP-1RA vs SGLT2i on risk of NAION, we additionally examined several optic disorders including diabetic retinopathy, macular degeneration, retinal vascular occlusion, and optic neuritis. We also assessed the frequency of ophthalmology or optometry visits during follow-up (eMethods in Supplement 1).
We reported the study following both the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and Transparent Reporting of Studies Emulating a Target Trial (TARGET) reporting guidelines. The study was approved with a waiver of informed consent by the VA St. Louis Health Care System institutional review board because the study involved minimal risk and could not be performed without the waiver. Data were analyzed using SAS Enterprise Guide version 8.3 (SAS Institute Inc).
Results
The study included 588 168 participants, with 139 546 who initiated GLP-1RAs (125 870 [90.20%] male, 27 394 [19.63%] Black, 106 829 [76.56%] White, and 5323 [3.81%] other races) and 448 622 who initiated SGLT2is (427 626 [95.32%] male, 90 876 [20.26%] Black, 338 624 [75.48%] White, and 19 122 [4.26%] other races). After weighting, age and BMI (calculated as weight in kilograms divided by height in meters squared) were similar between groups (mean [SD] age, 65.33 [10.98] vs 65.02 [10.95] years and mean [SD] BMI, 35.77 [7.01] vs 36.02 [7.12] for the GLP-1RA and SGLT2i groups, respectively). Baseline characteristics for the cohort are shown in the Table.
Table. Baseline Characteristics of Recipients of Glucagon-Like Peptide-1 Receptor Agonists (GLP-1RA) and Sodium-Glucose Cotransporter-2 Inhibitors (SGLT2i) Before and After Weighting.
| Baseline characteristics | Unweighted | Weighted | ||||
|---|---|---|---|---|---|---|
| Participants, No. (%) | SMDa | Participants, No. (%) | SMDa | |||
| GLP-1RA (n = 139 546) | SGLT2i (n = 448 622) | GLP-1RA (n = 139 546) | SGLT2i (n = 448 622) | |||
| Age, mean (SD), y | 65.33 (10.98) | 67.94 (10.68) | −0.24 | 65.33 (10.98) | 65.02 (10.95) | 0.03 |
| Sex | ||||||
| Male | 125 870 (90.20) | 427 626 (95.32) | −0.20 | 125 870 (90.20) | 402 015 (89.61) | 0.02 |
| Female | 13 676 (9.80) | 20 996 (4.68) | 0.20 | 13 676 (9.80) | 46 607 (10.39) | −0.02 |
| Raceb | ||||||
| Black | 27 394 (19.63) | 90 876 (20.26) | −0.02 | 27 394 (19.63) | 88 523 (19.73) | −0.003 |
| White | 106 829 (76.56) | 338 624 (75.48) | 0.03 | 106 829 (76.56) | 343 151 (76.49) | 0.002 |
| Otherc | 5323 (3.81) | 19 122 (4.26) | −0.02 | 5323 (3.81) | 16 948 (3.78) | 0.002 |
| Smoking status | ||||||
| Never | 56 901 (40.78) | 169 910 (37.87) | 0.06 | 56 901 (40.78) | 184 599 (41.14) | −0.01 |
| Former | 58 752 (42.10) | 194 250 (43.30) | −0.02 | 58 752 (42.10) | 186 972 (41.68) | 0.01 |
| Current | 23 893 (17.12) | 84 462 (18.83) | −0.04 | 23 893 (17.12) | 77 051 (17.18) | −0.001 |
| BMI, mean (SD)d | 35.77 (7.01) | 32.84 (6.30) | 0.44 | 35.77 (7.01) | 36.02 (7.12) | −0.04 |
| Estimated glomerular filtration rate, mean (SD), mL/min/1.73m2 | 75.01 (22.18) | 74.69 (20.41) | 0.02 | 75.01 (22.18) | 76.03 (21.75) | −0.05 |
| Systolic blood pressure, mean (SD), mmHg | 132.86 (16.56) | 133.71 (17.26) | −0.05 | 132.86 (16.56) | 132.83 (16.49) | 0.002 |
| Diastolic blood pressure, mean (SD), mmHg | 76.01 (9.84) | 76.14 (9.90) | −0.01 | 76.01 (9.84) | 76.02 (9.86) | −0.001 |
| HbA1c, mean (SD), % | 8.61 (2.43) | 8.23 (1.66) | 0.19 | 8.61 (2.43) | 8.65 (1.82) | −0.02 |
| HbA1c within 1 y, mean (SD) | 8.53 (1.67) | 8.13 (1.54) | 0.25 | 8.53 (1.67) | 8.58 (1.68) | −0.03 |
| Low-density lipoprotein, mean (SD), mg/dL | 83.49 (35.67) | 83.11 (35.86) | 0.01 | 83.49 (35.67) | 83.58 (35.74) | −0.002 |
| Cancer | 6257 (4.48) | 24 379 (5.43) | −0.04 | 6257 (4.48) | 19 751 (4.40) | 0.004 |
| AKI | 11 229 (8.05) | 28 408 (6.33) | 0.07 | 11 229 (8.05) | 35 004 (7.80) | 0.01 |
| Hyperlipidemia | 59 720 (42.80) | 179 982 (40.12) | 0.05 | 59 720 (42.80) | 193 219 (43.07) | −0.01 |
| Nonalcoholic fatty liver disease | 6983 (5.00) | 18 266 (4.07) | 0.05 | 6983 (5.00) | 23 726 (5.29) | −0.01 |
| Peripheral artery disease | 3003 (2.15) | 9850 (2.20) | −0.003 | 3003 (2.15) | 9884 (2.20) | −0.004 |
| Stroke | 3968 (2.84) | 14 609 (3.26) | −0.02 | 3968 (2.84) | 13 143 (2.93) | −0.01 |
| Myocardial infarction | 1544 (1.11) | 8134 (1.81) | −0.06 | 1544 (1.11) | 4907 (1.09) | 0.001 |
| Heart failure | 12 196 (8.74) | 59 167 (13.19) | −0.14 | 12 196 (8.74) | 36 988 (8.24) | 0.02 |
| Cataract | 39 566 (28.35) | 126 922 (28.29) | 0.001 | 39 566 (28.35) | 126 913 (28.29) | 0.001 |
| Macular degeneration | 8899 (6.38) | 31 933 (7.12) | −0.03 | 8899 (6.38) | 27 915 (6.22) | 0.01 |
| Diabetic retinopathy | 21 480 (15.39) | 52 403 (11.68) | 0.11 | 21 480 (15.39) | 69 451 (15.48) | −0.002 |
| Diabetic neuropathy | 43 544 (31.20) | 108 355 (24.15) | 0.16 | 43 544 (31.20) | 144 654 (32.24) | −0.02 |
| Diabetic nephropathy | 18 883 (13.53) | 48 583 (10.83) | 0.08 | 18 883 (13.53) | 58 269 (12.99) | 0.02 |
| Ocular surgery | 4428 (3.17) | 13 815 (3.08) | 0.01 | 4428 (3.17) | 13 851 (3.09) | 0.01 |
| Eye care visits or ophthalmology examination | 89 221 (63.94) | 271 669 (60.56) | 0.07 | 89 221 (63.94) | 288 732 (64.36) | −0.01 |
| Metformin | 93 123 (66.73) | 302 990 (67.54) | −0.02 | 93 123 (66.73) | 304 389 (67.85) | −0.02 |
| Insulin | 82 564 (59.17) | 146 494 (32.65) | 0.55 | 82 564 (59.17) | 271 963 (60.62) | −0.03 |
| DPP4i | 28 882 (20.70) | 84 283 (18.79) | 0.05 | 28 882 (20.70) | 100 401 (22.38) | −0.04 |
| Sulfonylureas | 46 024 (32.98) | 151 744 (33.82) | −0.02 | 46 024 (32.98) | 149 916 (33.42) | −0.01 |
| Thiazolidinediones | 8948 (6.41) | 22 430 (5.00) | 0.06 | 8948 (6.41) | 30 829 (6.87) | −0.02 |
Abbreviations: AKI, acute kidney injury; BMI, body mass index; DPP4i, dipeptidyl peptidase-4 inhibitors; HbA1c, hemoglobin A1c; SMD, standardized mean difference.
SI conversions: To convert HbA1c to proportion of total hemoglobin, multiply by 0.01; low-density lipoprotein to millimoles per liter, multiply by 0.0259.
SMD value range between −0.1 and 0.1 indicates balance was achieved.
Self-reported race information was collected from electronic health records and used in the study in accordance with the requirement by the funding agency (US Department of Veterans Affairs) and the Office of Management and Budget, which defines minimum standards for maintaining, collecting, and presenting data on race and ethnicity for all federal reporting agencies.
Other race included Alaska Native and American Indian, Asian or Native Hawaiian and Other Pacific Islander.
BMI calculated as weight in kilograms divided by height in meters squared.
Over 3 years of follow-up, the cumulative incidence of NAION was 39.07 (95% CI, 35.38-43.06) per 10 000 persons in the GLP-1RA group and 29.33 (95% CI, 26.34-33.06) per 10 000 persons in the SGLT2i group. Compared with SGLT2i, GLP-1RA use was associated with an increased risk of NAION (CID, 9.98; 95% CI, 3.48-14.03 per 10 000 persons at 3 years; CIR, 1.35; 95% CI, 1.11-1.51) (Figure).
Figure. Survival Curves Showing Cumulative Incidence of Nonarteritic Anterior Ischemic Optic Neuropathy (NAION) and Other Optic Disorders in Glucagon-Like Peptide-1 Receptor Agonists (GLP-1RA) and Sodium-Glucose Cotransporter-2 Inhibitors (SGLT2i) Groups.
A, Cumulative incidence of NAION; outcome defined as at least 1 International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code H47.01* diagnosis in any setting. Cumulative incidence per 10 000 persons for GLP-1RA and SGLT2i; cumulative incidence difference and cumulative incidence ratio at 3 years are included. B through D, Cumulative incidence of alternative NAION definitions. B, Outcome defined by at least 1 ICD-10 code H47.01* made by an ophthalmology or optometry specialist. C, Outcome defined by at least 2 NAION diagnoses more than 30 days apart in any clinical setting. D, Outcome defined by at least 2 NAION diagnoses more than 30 days apart, at least 1 of which was made by an ophthalmology or optometry specialist. E through H, Cumulative incidence of other optic disorders. I, frequency of ophthalmology and optometry visit.
The increased risk associated with GLP-1RA use was consistently observed across alternative NAION definitions, including NAION diagnosis by an eye-care specialist (CID, 8.73; 95% CI, 2.46-12.89; CIR, 1.34; 95% CI, 1.08-1.52), repeated NAION diagnoses (CID, 6.35; 95% CI, 2.40–9.65; CIR, 1.46; 95% CI, 1.14-1.76), and specialist diagnosed NAION with repeated diagnosis (CID, 5.91; 95% CI, 2.00-8.88; CIR, 1.43; 95% CI, 1.13-1.74) (Figure). We additionally evaluated other optic disorders to assess the specificity of the observed association with NAION. Compared with SGLT2i use, GLP-1RA use was not associated with risk of other ophthalmic disorders, including diabetic retinopathy (CIR, 1.01; 95% CI, 0.99-1.03), macular degeneration (CIR, 1.00; 95% CI, 0.97–1.05), retinal vascular occlusion (CIR, 0.98; 95% CI, 0.89-1.05), or optic neuritis (CIR, 1.00; 95% CI, 0.77-1.32). The frequency of ophthalmology or optometry clinic visits during follow-up was also similar between groups (CIR, 1.00; 95% CI, 0.99-1.00) (Figure).
Discussion
In this large cohort study, GLP-1RA use was associated with a modestly increased risk of NAION compared with SGLT2i use. While the absolute risk remains low, the specificity of this finding may warrant heightened vigilance.
This study has limitations. The cohort was older and predominantly male. Residual confounding, selection bias, and outcome misclassification, including incomplete capture of events outside the VA, cannot be fully excluded.
eMethods.
eReferences.
Data Sharing Statement
References
- 1.Xie Y, Choi T, Al-Aly Z. Mapping the effectiveness and risks of GLP-1 receptor agonists. Nat Med. 2025;31(3):951-962. doi: 10.1038/s41591-024-03412-w [DOI] [PubMed] [Google Scholar]
- 2.Cai M, Choi T, Xie Y, Al-Aly Z. Glucagon-like peptide-1 receptor agonists and risk of substance use disorders among US veterans with type 2 diabetes: cohort study. BMJ. 2026;392:e086886. doi: 10.1136/bmj-2025-086886 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Zhou J, Huang W, Xie Y, Shen H, Liu M, Wu X. Risk of ophthalmic adverse drug reactions in patients prescribed glucagon-like peptide 1 receptor agonists: a pharmacovigilance study based on the FDA adverse event reporting system database. Endocrine. 2025;88(1):80-90. doi: 10.1007/s12020-024-04112-8 [DOI] [PubMed] [Google Scholar]
- 4.Massy M, Marti S, Hammer H, Hoepner R. Increased vision impairment reports linked to semaglutide: analysis of FDA adverse event data. BMC Med. 2025;23(1):203. doi: 10.1186/s12916-025-04031-z [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hathaway JT, Shah MP, Hathaway DB, et al. Risk of nonarteritic anterior ischemic optic neuropathy in patients prescribed semaglutide. JAMA Ophthalmol. 2024;142(8):732-739. doi: 10.1001/jamaophthalmol.2024.2296 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Abbass NJ, Nahlawi R, Shaia JK, et al. The effect of semaglutide and GLP-1 RAs on risk of nonarteritic anterior ischemic optic neuropathy. Am J Ophthalmol. 2025;274:24-31. doi: 10.1016/j.ajo.2025.02.025 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods.
eReferences.
Data Sharing Statement
