Your phone lights up. "Ozempic linked to blindness." Your stomach drops, because maybe the pen is in your fridge right now, or your sister just started, or you were finally thinking about asking your doctor. Before that headline decides how you feel, give it sixty seconds and three questions.
No science degree needed. This is a short, repeatable check that hands the judgment back to you, so the next GLP-1 headline lands as information instead of a punch. We will use the newest one as the example: the eye warning.
On 9 November 2026 the U.S. Food and Drug Administration published a drug safety communication. It requires the makers of all GLP-1 medications, including Ozempic, Wegovy, Rybelsus, Mounjaro and Zepbound, to add a warning to the label about a potential increased risk of a rare eye condition called NAION (non-arteritic anterior ischemic optic neuropathy).
In plain words: NAION happens when blood flow to the front of the optic nerve drops suddenly. It usually affects one eye. The FDA describes it as sudden, painless vision loss in one eye that people may notice when they wake up. It is rare: the FDA puts the usual rate at roughly 2 to 10 cases per 100,000 adults over 50 each year. In people with type 2 diabetes the risk is higher to begin with. In the study below it was about 5.5 per 10,000 over 18 months without a GLP-1, and about 3 more with one.
Look at the word the FDA chose: potential. The agency wrote that it found a potential increased risk of rare NAION events, and it said its review of animal data did not find a mechanism that would explain one. That word is your first clue that the story is less settled than the headline sounds.
Your move: next time a GLP-1 headline scares you, find the original wording first (the FDA page, the study title, the abstract). Copy the exact verb. "Causes", "linked to" and "potential" are three very different claims.
Much of the coverage leaned on a large study published in July 2026 in the Annals of Internal Medicine by a Rutgers team, Reynolds and colleagues. They followed U.S. adults with type 2 diabetes who started a GLP-1 medication or one of two other diabetes medications, and counted eye events over 18 months.
Here is what they found, compared with people starting an SGLT2 inhibitor (another diabetes drug): 8.5 cases per 10,000 people on a GLP-1, and 5.5 cases per 10,000 on the comparison drug. Same study, same numbers. Now watch two honest headlines come out of them.
| The headline you see | Type of number | How it is calculated | What it tells you |
|---|---|---|---|
| "GLP-1 users had a 55% higher risk" | Relative risk | 8.5 divided by 5.5 is about 1.55 | How the two groups compare to each other. Says nothing about how common the event is. |
| "About 3 extra cases per 10,000 people" | Absolute risk difference | 8.5 minus 5.5 is 3.0 | How many more people out of 10,000 had the event over 18 months. |
| "More than 9,990 of 10,000 had no event" | The same absolute number, flipped | 10,000 minus 8.5 | What happened to almost everyone in the GLP-1 group during the study. |
| "About 1 extra case for every 3,300 people treated" | Number needed to harm | Reported by the authors (3,333) | How many people would need to take the drug for one extra case, if the link is real. |
None of these numbers is a lie. A relative number tells you the direction. The absolute number tells you the size. A rare event can rise by 55% and still stay rare. That is exactly why the authors themselves wrote that the absolute risk "remained very low".
Both numbers matter. If you already carry several risk factors, a small absolute difference can still be worth a conversation with your doctor. The point of the check is proportion, so you can decide how much space this headline gets in your head.
Your move: whenever you see a percentage in a health headline, ask "percent of what?" Then look for the plain count per 1,000 or per 10,000 people. If the article never gives it, that is information too.
The protocol
The 60-second number check
Use it on any GLP-1 headline: eyes, pancreas, muscle, mood, whatever comes next. You need the article and, ideally, one click through to the study or the official statement. Rule: no decision about your medication from a headline, only a question for your doctor.
Save or screenshot this
| Ask | Where to look | The eye warning, checked |
|---|---|---|
| 1. Who was studied? | "Participants" or "patients" in the abstract | Adults aged 18 to 65 with type 2 diabetes. In the GLP-1 group, most eye events happened in people over 50 and in men. |
| 2. Relative or absolute? | The results line: per 1,000 or per 10,000, "risk difference" | 55% higher (relative) equals about 3 extra cases per 10,000 over 18 months (absolute). |
| 3. Potential or proven? | The verbs and the limitations paragraph | "Potential" and "associated with". Enough signal to add a warning, proof of cause still thin. |
This is the question most headlines skip. A study can be careful, large and true for the people in it, and still tell you little about yourself. The Rutgers study looked at people with type 2 diabetes. The authors noted that the differences were bigger in men, in people aged 50 or older, and in people with heart disease or existing eye conditions, and minimal in women and in people under 50.
| Check | In the Rutgers study | Me |
|---|---|---|
| Why the medication is taken | Type 2 diabetes | Diabetes, weight, both, or not taking it |
| Age | 18 to 65; in the GLP-1 group, most eye events in people over 50 | Under 50 or 50 and over |
| Sex | About 7 in 10 eye events in the GLP-1 group were in men | |
| Other health factors | Bigger differences with heart disease, eye conditions or two or more diabetes medications | Any of these apply to me? |
What about people who take a GLP-1 for weight? The evidence there is thinner. A 2026 review in Ophthalmology that split the data by reason for use found no clear increase in NAION in people taking semaglutide for weight loss, in observational studies or in randomised trials. The numbers were small, so this is "no clear signal so far" rather than proof of safety. "Not studied in people like me" is a different answer from "risky for me" and from "safe for me".
Your move: fill in the "Me" column above. Every row where you and the study differ is a question to bring to your doctor, not a reason to panic or to relax.
A drug label warning does not mean the question is closed. Regulators add a warning when there is enough signal that people deserve to know, even while the proof is still thin. In this case the FDA pulled together reports, a pooled analysis of clinical trials and observational studies. The studies point in a similar direction, and most of them share the same weak spots.
| Wording you see | What it usually means | How much weight to give it |
|---|---|---|
| "Causes" | A cause has been shown, usually by randomised trials | Check whether the study design can actually show that |
| "Linked to" or "associated with" | Two things showed up together in the data | A real signal, cause not shown |
| "Potential risk" | A signal regulators take seriously enough to warn about | Worth knowing and asking about, still open |
| "May reflect residual confounding" | Something else about the people could explain part of it | The authors are telling you to stay cautious |
That last row is not my opinion. The Rutgers authors and a large Swedish study published in the same issue both wrote that the differences they saw may reflect residual confounding. People who are prescribed a GLP-1 for diabetes can differ in many ways from people prescribed something else, for example in how long they have had diabetes or how high their blood sugar is, and those differences can affect the eyes too.
"This deserves a warning" and "this is not proven yet" can both be true. That is what honest science looks like while the evidence is still coming in.
If you take a GLP-1 medication, the most important thing is what you don't do: never stop or start a medication on your own because of a headline. Stopping and starting are decisions for you and your doctor together. The 2026 consensus statement from the North American Neuro-Ophthalmology Society and the American Academy of Ophthalmology says exactly this: whether to start, continue or stop is a shared decision between each patient and their care team.
If you notice a sudden change in your vision, especially in one eye, get checked the same day. That applies whether or not you take any medication. The FDA asks people to contact their health care professional right away with these symptoms.
For the conversation with your doctor, three sentences are enough:
- "I read about the eye warning for GLP-1 medications. Does it apply to someone like me?"
- "Do I have any of the risk factors where the difference was bigger, like heart disease or an existing eye condition?"
- "Is there anything I should watch for, and would an eye check make sense for me?"
Not taking a GLP-1 but considering it? Same logic. Put the headline on your list of questions. It is one piece of information next to everything else you and your doctor weigh.
For the bigger picture on what these medications do and don't do for eating, read The GLP-1 Reality Check. It covers food noise, what happens after you stop, and why habits still matter with or without the drug.
Free, straight to your inbox
Get the 60-second number check card
A one-page card with the three questions and a "does this study look like me?" box to keep on your phone for the next GLP-1 headline, plus one new guide from me every week. Real research, real life, no diet rules.
No spam. Unsubscribe with one click, any time.
If you live with obesity
This guide is a building block, not a treatment
Obesity is a chronic disease, not a lack of discipline. It has many causes, from genes and hormones to sleep, stress and medication. The European Association for the Study of Obesity describes it as a complex, relapsing, chronic disease, and a 2025 Lancet commission defined clinical obesity as a chronic illness that deserves fair access to evidence-based treatment.
There are effective treatments: medical care, nutrition therapy, exercise therapy, medication and, for some people, surgery. The earlier you start, the more options you have.
If you are living with obesity, please talk to your doctor or an obesity specialist and get into the right hands. You deserve real support, not just another set of rules. Use this guide alongside your care, as one small step that is yours.
What the research actually shows, and where each finding stops
Five sources sit behind this guide. Each one also has an edge where its evidence stops, worth knowing before you lean on it.
The label change. The FDA drug safety communication of 9 November 2026 requires all GLP-1 medications to carry a warning about a potential increased risk of NAION. It is based on adverse event reports, a pooled analysis of 126 clinical trials (about 6.4 versus 3.8 cases per 100,000 person-years) and published observational studies. The edge: the FDA does not state that GLP-1 medications cause NAION, found no mechanism in animal data, and notes that NAION risk factors overlap with conditions common in people who take these drugs.
The headline numbers. A 2026 target trial emulation in the Annals of Internal Medicine by Reynolds and colleagues used U.S. insurance claims (2017 to 2022) from adults aged 18 to 65 with type 2 diabetes. Over 18 months, the risk of ischemic optic neuropathy was 8.5 versus 5.5 per 10,000 compared with SGLT2 inhibitors (difference 3.0) and 7.8 versus 4.2 per 10,000 compared with DPP4 inhibitors (difference 3.6). Funded by the National Institutes of Health. The edge: observational data; there was no diagnosis code specifically for NAION, so a broader eye diagnosis was used as a stand-in; body weight and diabetes duration were missing; the authors say the difference may reflect residual confounding. Results apply to adults with type 2 diabetes, not to people taking GLP-1s for weight.
The same question in Sweden. A 2026 nationwide cohort study in the Annals of Internal Medicine by Ueda and colleagues compared about 107,500 GLP-1 users with about 185,900 SGLT2 inhibitor users with type 2 diabetes. At one year the risk was 0.04% versus 0.02%. When the analysis was limited to people already on metformin, to compare more similar patients, the difference shrank considerably. The edge: few events overall (62 and 64), possible unmeasured confounding, and the authors say it generalises poorly to people without diabetes.
People taking it for weight. A 2026 systematic review and meta-analysis in Ophthalmology by Liu and colleagues pooled 8 randomised trials and 8 observational studies and separated diabetes from weight loss. The risk signal appeared in observational data in diabetes only. For weight loss, observational studies showed no evidence of increased NAION hazards, and trials found 4 events versus 1 without a significant difference. The edge: very few events and wide uncertainty, so this cannot rule out a small risk; it is about semaglutide only; disclosures are listed in the full article, not the abstract.
How headlines get it wrong. A 2026 content analysis in the Journal of Neuro-Ophthalmology by Mahiny and colleagues reviewed 71 English-language news articles on semaglutide and NAION. About half (48%) contained misleading claims and 21% used causal language. Articles that avoided causation errors more often quoted physicians and explained the study's limits. The edge: one news cycle (2024 to 2025), English-language media only, and coding by the study team, so it describes patterns rather than measuring harm to readers.
For clinical guidance, the 2026 consensus statement by the North American Neuro-Ophthalmology Society and the American Academy of Ophthalmology notes that the evidence is mostly retrospective, that some studies find no link, that the absolute risk remains low, and that start, continue or stop is a shared decision. The edge: expert opinion, not new data.
The takeaway
A scary GLP-1 headline is a reason to ask three questions before you change anything. Who was studied, relative or absolute, potential or proven. Sixty seconds, and the number is back in proportion. Then the real decision goes where it belongs: to you and your doctor.
Which GLP-1 headline made your stomach drop most recently, and what happens when you run it through the three questions?
