Sudden painless vision loss in one eye is the symptom behind this study, and the finding is real enough to have changed labeling conversations worldwide. A retrospective matched cohort at one academic center searched 16,827 patients with no history of the condition and compared those prescribed semaglutide against those prescribed other treatments[1]. The hazard ratios are large. The population they came from is the part that decides what they mean for anyone reading a seller’s page.
What was found
Among 710 patients with type 2 diabetes, 17 events of nonarteritic anterior ischemic optic neuropathy occurred among those prescribed semaglutide against 6 in the comparator group. Cumulative incidence over 36 months was 8.9 per cent (95% CI 4.5 to 13.1) against 1.8 per cent (95% CI 0 to 3.5), a hazard ratio of 4.28 (95% CI 1.62 to 11.29, P < 0.001).
Among 979 patients who were overweight or obese, 20 events occurred on semaglutide against 3 on the comparator. Cumulative incidence was 6.7 per cent (95% CI 3.6 to 9.7) against 0.8 per cent (95% CI 0 to 1.8), a hazard ratio of 7.64 (95% CI 2.21 to 26.36).
Where those people came from
Every patient in this study was evaluated by neuro-ophthalmologists at a single academic institution. That is a referral population: people already sent to specialists for eye problems. A cumulative incidence of 6.7 per cent in that group says nothing like 6.7 per cent for a general population, because the denominator has been filtered by the exact thing being measured.
The interval widths say the same thing more quietly. A hazard ratio of 7.64 with a confidence interval running from 2.21 to 26.36 is built on 23 events in total. The direction is consistent and the magnitude is unsettled, which is what a large interval on small counts looks like.
What to do with it
NAION is uncommon and it is not reversible, so a small absolute risk is still worth knowing. The practical response is not to avoid the drug on the strength of one referral cohort, and it is not to dismiss the finding either. It is to treat sudden painless vision loss in one eye as an emergency, and to tell a clinician about it the same day rather than at the next refill.
That is harder than it sounds when the prescription arrived through a website. Most of the sellers tracked here publish nothing about who to contact or how quickly, which is the gap we count in what sellers will not tell you and list in the questions they leave open. Knowing the route to a human before you need one is the part of this a buyer can actually control, and the disclosure census shows who publishes it.