GLP-1s and your eyes: the NAION safety signal, explained

In mid-2024, a study out of Massachusetts Eye and Ear put a specific, rare eye condition — NAION — on the map for GLP-1 users. It's since been examined in a much larger European registry study. Here's what both actually found, with the real numbers rather than a vague "may increase risk" summary.

What NAION is

Non-arteritic anterior ischemic optic neuropathy (NAION) is a sudden, typically painless loss of vision in one eye, caused by reduced blood flow to the optic nerve. It's uncommon in the general population, tends to occur in people over 50 with vascular risk factors like diabetes and hypertension, and any vision loss it causes is usually permanent — there's no established treatment that reliably reverses it once it occurs. That severity, combined with how it typically strikes without warning, is why the signal below drew immediate clinical attention.

The Harvard/Mass Eye and Ear study

A retrospective matched cohort study compared NAION incidence in semaglutide users against non-GLP-1 users, separately in patients with type 2 diabetes and in patients with overweight or obesity¹. Among patients with diabetes, the cumulative NAION incidence over 36 months was 8.9% in semaglutide users versus 1.8% in the comparison group — a hazard ratio of 4.28. Among patients with overweight or obesity, the incidence was 6.7% versus 0.8%, a hazard ratio of 7.64. Both associations were statistically significant. The authors were explicit that this was an observational study and could not establish that semaglutide causes NAION, only that it found a strong association.

The Danish-Norwegian registry study

A much larger follow-up study used national health registries across Denmark and Norway, comparing over 61,000 semaglutide users against SGLT-2 inhibitor users (a different diabetes drug class) as the comparison group, with 32 total NAION events observed². The pooled adjusted hazard ratio was 2.81 — lower than the Harvard study's figures, but still a statistically significant, real association. Critically, the authors emphasized that the absolute risk remained low: the incidence rate difference was about 1.41 additional NAION cases per 10,000 person-years, meaning the vast majority of semaglutide users in this cohort did not develop NAION.

Why the two studies don't perfectly agree

A hazard ratio of 4.28–7.64 versus 2.81 is a real, notable difference, and it's worth naming honestly rather than picking whichever number sounds more or less alarming. Different comparison groups (non-GLP-1 users broadly versus SGLT-2 inhibitor users specifically), different populations, and different sample sizes can all shift a hazard ratio without either study being wrong. What both studies agree on is the direction and statistical significance of the association — semaglutide users had a meaningfully higher rate of NAION than the comparison group in both cases — even as the exact magnitude differs. The original authors themselves responded to follow-up commentary reinforcing that this remains a genuine safety signal warranting further study and clinical awareness, not a settled causal mechanism³.

What this means practically

NAION remains rare in absolute terms even with an elevated relative risk, so this isn't a reason for most people to avoid GLP-1 therapy outright. It is a reason to disclose any personal or family history of NAION, existing optic nerve conditions, or significant uncontrolled vascular risk factors to your prescriber before starting — the same kind of real intake screening covered in how to choose a GLP-1 provider — and to treat any sudden vision change in one eye, however mild it seems, as an urgent same-day medical issue rather than something to monitor at home.

The honest bottom line

Two independent studies, using different methods and populations, both found a statistically significant association between semaglutide and NAION, with hazard ratios ranging from roughly 2.8 to 7.6 depending on the comparison group — a real signal, not a single outlier finding. The absolute risk remains low, and neither study proves direct causation, but both are consistent enough that sudden one-eye vision loss deserves urgent attention in anyone on a GLP-1. This is educational information, not medical advice; seek immediate care for any sudden vision change.