FDA Data Reveal Different Risks for GLP-1s

A new analysis of the FDA’s adverse event database is challenging the idea that all GLP-1 medications carry the same risks. The study, published in Obesity and reported by Medscape, found that safety profiles vary significantly depending on which drug you’re taking and why you’re taking it. For the millions of people now on these medications, that distinction matters a lot.

What the researchers found

The team analyzed reports from the FDA’s Adverse Event Reporting System (FAERS) spanning 2012 to 2025. They looked at five of the most commonly prescribed GLP-1 receptor agonists: exenatide, liraglutide, dulaglutide, semaglutide, and tirzepatide. What emerged wasn’t a single risk profile but several, split along lines that most patients probably haven’t considered.

The biggest surprise? The reason you’re prescribed a GLP-1 appears to shape your risk exposure.

When used for type 2 diabetes, GLP-1s were associated with:

• Retinopathies (eye damage)
• Skin and subcutaneous tissue conditions
• Pancreatic neoplasms
• Hearing loss
• Cataracts

When used for weight management, a different pattern emerged:

• Nutritional disorders
• Sensory abnormalities
• Panic attacks and panic disorders
• Suicidal or self-injurious behavior
• Depressive disorders
• Immune-associated conditions
• Eating disorders

That’s a striking split. Diabetes patients saw more reports involving eyes, skin, and the pancreas. Weight-loss patients saw more psychiatric and neurological signals. Whether this reflects the drugs themselves, the underlying patient populations, or some combination of both is something researchers will need to untangle.

Not all GLP-1s are created equal

The data also showed clear differences between individual drugs. Semaglutide, the active ingredient in Ozempic and Wegovy, was associated with more gastrointestinal and psychiatric adverse events. Dulaglutide (Trulicity) and exenatide (Byetta, Bydureon) showed higher rates of deaths and hospitalizations in reports. Liraglutide (Victoza, Saxenda) and exenatide had more neoplasm-related reports.

This doesn’t mean one drug is definitively “safer” than another. FAERS data has well-known limitations. It’s a voluntary reporting system, not a controlled trial. Reporting rates can be influenced by media coverage, market share, and how long a drug has been available. Semaglutide’s outsized media presence, for instance, could drive more reports simply because more people are paying attention to their symptoms.

Still, the pattern is consistent enough to warrant a harder look.

A signal on women’s health

Among female patients specifically, GLP-1 use was linked to reports of menstrual bleeding irregularities, ovarian and fallopian cysts and neoplasms, and reproductive system hemorrhages. This finding adds to a growing conversation about how these drugs affect women differently, particularly as more women of reproductive age start using them for weight loss.

It’s an area where clinical data has lagged behind real-world prescribing patterns. These FAERS signals don’t prove causation, but they point to questions that clinical trials should be designed to answer.

Why this matters right now

The GLP-1 class has exploded. Tens of millions of prescriptions are being written annually, and the patient population has shifted dramatically from primarily diabetic to a much broader group seeking weight management. That shift changes the risk calculus.

If you’re a diabetes patient, your doctor is weighing GLP-1 side effects against the very real dangers of uncontrolled blood sugar. If you’re an otherwise healthy person looking to lose 20 pounds, the acceptable risk threshold is different. This study suggests the actual risks may be different too.

The practical takeaway: talk to your prescriber about which specific GLP-1 makes sense for your situation, not just whether to take one. These drugs aren’t interchangeable, and the emerging data says their risk profiles aren’t either. As more targeted safety data becomes available, expect prescribing to get more personalized. That’s a good thing.