Data and Timing-Related Concerns About Semaglutide and NAION Risk

By
Chun-Hsien Yu
October 1, 2026

Jama Ophthalmology

At a Glance

Category	Detail
Condition	Nonarteritic anterior ischemic optic neuropathy (NAION) risk associated with semaglutide initiation
Key Mechanisms	The letter raises concerns about unexplained changes in the risk set, possible informative censoring or data-cleaning artifacts, nonproportional hazards, and potential timing of risk relative to treatment initiation or rapid glycemic shifts.
Target Population	US veterans initiating semaglutide or sodium-glucose cotransporter-2 inhibitors (SGLT2i)
Care Setting	Clinical interpretation and counseling about NAION risk during treatment initiation and follow-up

Key Highlights

The letter questions a risk-set decrease of 3,748 SGLT2i initiators by day 182.5, while reported NAION events and censoring account for 545 individuals.
The authors of the letter calculate that 3,203 patients, approximately 3.5% of the cohort, are not accounted for in the risk pool during the first 6 months.
The Kaplan-Meier curves reportedly begin to separate at approximately 1 year, raising concern that a single hazard ratio may not represent risk consistently over time.
The reported hazard ratio is 2.33; the letter asks whether it differs between the first treatment year and subsequent follow-up.
The letter notes reliance on administrative coding and cites an approximately 74.5% positive predictive value for NAION.

Guideline-Based Recommendations

Diagnosis

The letter identifies administrative coding as the basis for NAION ascertainment and cites an approximately 74.5% positive predictive value.

Management

The letter states that clarifying data-accounting and temporal-risk concerns is important for clinicians assessing NAION risk during patient counseling.

Monitoring & Follow-up

The letter asks whether the hazard ratio differs between the first treatment year and ensuing follow-up; it does not report an answer.

Risks

The letter raises the possibility that unexplained risk-set attrition could reflect informative censoring or a data-cleaning artifact and could bias the reported hazard ratio.
The reported separation of Kaplan-Meier curves at approximately 1 year raises concern that a single hazard ratio may obscure temporal variation in risk.

Patient & Prescribing Data

US veterans initiating semaglutide or SGLT2i in the study discussed by the letter.

The letter discusses a reported hazard ratio of 2.33 for NAION and describes the risk as rare but evident; it raises questions about risk-set accounting and timing of risk rather than providing prescribing recommendations.

Clinical Best Practices

When interpreting the reported association, consider the letter's concern that the SGLT2i risk-set decrease is not fully explained by reported events and censoring.
Consider the reported timing of Kaplan-Meier curve separation when interpreting a single hazard ratio.
The letter identifies accurate cohort accounting and assessment of time-varying hazard ratios as unresolved questions.