Background
Semaglutide is a glucagon-like peptide-1 receptor agonist, or GLP-1RA, widely used to improve blood sugar control in adults with type 2 diabetes and, in some settings, to support weight management. Because GLP-1RAs can produce rapid improvements in glycemic control and may influence vascular biology, questions have been raised about whether semaglutide could affect eye health, including the risk of neovascular age-related macular degeneration, or NVAMD.
NVAMD is an advanced form of age-related macular degeneration in which abnormal blood vessels grow under the macula, the central part of the retina responsible for sharp vision. These fragile vessels can leak fluid or blood, leading to distortion of vision and sometimes severe central vision loss. Although NVAMD is not a classic complication of diabetes, it is a major retinal disease with important public health implications, especially in older adults.
This study examined whether semaglutide use in adults with type 2 diabetes was associated with a higher or lower risk of NVAMD compared with other common glucose-lowering drugs.
Study Design
This was a retrospective network study conducted across 12 databases within the Observational Health Data Sciences and Informatics, or OHDSI, network. The study period ran from December 1, 2017 through December 31, 2024.
Researchers focused on adults with type 2 diabetes who newly started one of several medications:
Semaglutide
Other GLP-1 receptor agonists, specifically dulaglutide and exenatide
Non-GLP-1 comparator drugs, including empagliflozin, sitagliptin, and glipizide
Two analytic approaches were used to improve confidence in the findings. First, an active-comparator cohort design compared people starting semaglutide with people starting another medication used for diabetes. Second, a self-controlled case-series, or SCCS, analysis examined whether the timing of semaglutide exposure was associated with NVAMD risk within the same person over time.
Using both methods helps reduce different kinds of bias. The cohort approach compares groups of patients, while the SCCS approach controls for fixed characteristics within each person, such as genetics, long-standing health habits, and many baseline risk factors.
Outcome Definitions
The main outcome was NVAMD, defined in two ways:
NVAMD-C: based on diagnosis or condition codes alone
NVAMD-CP: based on diagnosis codes plus procedure codes, creating a stricter definition that may better capture clinically confirmed disease
Using two definitions allowed the investigators to test whether results were consistent when the outcome definition became more specific.
Results
The study included 227,971 new users of semaglutide, making it a very large real-world evaluation.
When semaglutide was compared with dulaglutide, no significant difference in NVAMD risk was seen:
NVAMD-C hazard ratio, 0.57; 95% confidence interval, 0.21 to 1.57
NVAMD-CP hazard ratio, 0.25; 95% confidence interval, 0.05 to 1.27
Compared with empagliflozin, semaglutide also showed no meaningful difference:
NVAMD-C hazard ratio, 0.98; 95% confidence interval, 0.54 to 1.79
NVAMD-CP hazard ratio, 0.79; 95% confidence interval, 0.38 to 1.64
Compared with sitagliptin, the point estimate was somewhat higher for NVAMD-C, but the results were not statistically significant and the confidence intervals were wide:
NVAMD-C hazard ratio, 2.08; 95% confidence interval, 0.90 to 4.83
NVAMD-CP hazard ratio, 1.80; 95% confidence interval, 0.55 to 5.86
Compared with glipizide, no significant difference was found:
NVAMD-C hazard ratio, 0.83; 95% confidence interval, 0.35 to 2.02
NVAMD-CP hazard ratio, 0.50; 95% confidence interval, 0.21 to 1.19
The self-controlled case-series analysis also found no evidence that semaglutide increased or decreased the incidence of NVAMD over time:
NVAMD-C incidence rate ratio, 0.92; 95% confidence interval, 0.67 to 1.26
NVAMD-CP incidence rate ratio, 1.02; 95% confidence interval, 0.76 to 1.36
Likewise, no clear change in risk was observed for the other GLP-1RA or non-GLP-1RA comparators.
Importantly, the confidence intervals in several comparisons were wide, which means the study did not identify a precise protective or harmful effect. Rather, the overall pattern was neutral: no convincing signal that semaglutide changes NVAMD risk in adults with type 2 diabetes.
Clinical Interpretation
These findings are reassuring for clinicians and patients using semaglutide for type 2 diabetes. Based on the available data from this large network study, semaglutide does not appear to alter the risk of developing neovascular age-related macular degeneration.
That said, the absence of a detected association does not mean semaglutide has no eye-related considerations at all. In diabetes care, rapid improvement in blood glucose can sometimes temporarily worsen diabetic retinopathy in susceptible patients, especially when glucose control changes quickly. This phenomenon is distinct from NVAMD, which has a different underlying cause and is not the same as diabetic retinopathy or diabetic macular edema.
Patients with diabetes should still receive routine eye examinations according to current guidelines, because diabetes itself remains a major risk factor for several retinal diseases. Older adults should also be monitored for age-related macular degeneration based on standard ophthalmic practice, especially if they develop blurry central vision, straight lines appearing wavy, dark spots, or difficulty reading.
Strengths and Limitations
This study has several strengths. It included a very large population from multiple databases, used two complementary analytic methods, and tested two outcome definitions. These features improve robustness and help reduce the chance that the findings are due to a single database, a single method, or a single case definition.
However, as with all observational research, there are limitations. The study cannot prove causation. Residual confounding is possible, meaning unmeasured factors may still influence the results. Administrative data may also miss some clinical details, such as smoking history, visual acuity, disease severity, retinal imaging findings, or reasons a clinician chose one diabetes drug over another.
In addition, NVAMD is relatively uncommon, so even in a large dataset the number of events may still be limited, especially within certain comparison groups. This can lead to wide confidence intervals and reduced precision.
Bottom Line
In this OHDSI network study of adults with type 2 diabetes, semaglutide was not associated with an increased or decreased risk of neovascular age-related macular degeneration. The findings support ocular safety with respect to NVAMD and are helpful for clinicians weighing diabetes treatment options.
Patients taking semaglutide should continue regular diabetes and eye follow-up, but this study does not suggest a specific need to avoid semaglutide because of concern for NVAMD alone.