What the safety studies on GLP-1 drugs found

Every few months a new study links GLP-1 drugs to a serious harm, and a few months later another study finds no link. Most of the confusion comes from mixing two kinds of evidence. A randomised trial assigns the drug by chance, so a difference in harm between groups is likely caused by the drug, but trials are often too small or short to catch rare events. An observational study looks at people who chose or were prescribed the drug in practice. It can be huge, but it shows association, not cause, because the people on the drug differ from those who are not. Each finding below is labelled. The everyday side effects are on the side effects page.

Gallbladder: randomised, and consistent

This is the clearest signal, and it comes from randomised data. A 2022 meta-analysis in JAMA Internal Medicine pooled 76 trials with 103,371 participants. GLP-1 drugs raised the risk of gallbladder or biliary disease by 37 percent (relative risk 1.37, 95% CI 1.23 to 1.52), including gallstones (RR 1.27) and cholecystitis (RR 1.36). The risk was higher in weight-loss trials (RR 2.29), at higher doses and with longer use. Rapid weight loss itself raises gallstone risk, which is part of the explanation. A relative increase on a low baseline still means a small absolute risk for most people.

Pancreatitis: trials say no, one cohort says yes

Pancreatitis is on every GLP-1 label. The randomised evidence has not shown an increase. A 2017 meta-analysis of three long-term placebo-controlled trials in type 2 diabetes, with 9,347 people on a GLP-1 drug and 9,353 on placebo and every case adjudicated blind, found an odds ratio of 0.745 (95% CI 0.47 to 1.17). The authors noted more evidence was needed.

The study most quoted on the other side is a 2023 JAMA research letter by Sodhi and colleagues. It used US insurance claims from 2006 to 2020 and compared 4,144 people on liraglutide and 613 on semaglutide for weight loss with 654 on bupropion-naltrexone. The adjusted hazard ratios were 9.09 for pancreatitis (95% CI 1.25 to 66.00), 4.22 for bowel obstruction (1.02 to 17.40) and 3.67 for gastroparesis (1.15 to 11.90). Biliary disease was not significantly raised (1.50, 0.89 to 2.53). The intervals are very wide because the events were few: there was one case of pancreatitis in the comparison group and two among the semaglutide users. It is a signal worth knowing about, and an observational one.

Thyroid cancer: two cohorts, two answers

The boxed warning comes from rodent studies, where these drugs caused C-cell tumours. In people, the evidence is observational and mixed. A French nested case-control study in Diabetes Care (2023) matched 2,562 people with thyroid cancer to 45,184 controls and found that one to three years of GLP-1 use was associated with a higher risk of all thyroid cancer (HR 1.58, 95% CI 1.27 to 1.95) and of medullary thyroid cancer (HR 1.78, 1.04 to 3.05).

A larger Scandinavian cohort in the BMJ (2024) compared 145,410 people starting a GLP-1 drug with 291,667 starting a DPP-4 inhibitor, a design chosen to reduce confounding. Over a mean 3.9 years there was no increased risk (HR 0.93, 95% CI 0.66 to 1.31). For medullary thyroid cancer the HR was 1.19, with an interval from 0.37 to 3.86 that is too wide to rule much in or out. The authors concluded the data were consistent with no more than a 31 percent relative increase. The label contraindication for a personal or family history of medullary thyroid carcinoma or MEN 2 is unchanged.

Suicidal thoughts: reviewed and cleared

After case reports from Iceland in 2023, both main regulators reviewed the question. The EMA's safety committee, PRAC, concluded in April 2024 that the evidence did not support a causal link and that no label change was needed. The FDA's preliminary review in January 2024 found no evidence of a link but said it could not rule out a small risk. In January 2026 the FDA completed that work: a meta-analysis of 91 placebo-controlled trials with 107,910 participants (60,338 on a GLP-1 drug, 47,572 on placebo) found no increased risk of suicidal ideation or behaviour, or of anxiety, depression or psychosis, and a Sentinel claims study of more than 2.2 million people with type 2 diabetes found no increase in intentional self-harm compared with SGLT2 inhibitors. On 13 January 2026 the FDA asked for the suicidality warning to be removed from the Wegovy, Zepbound and Saxenda labels. A change in mood is still worth raising with a prescriber, whatever the cause.

NAION: a very rare eye side effect of semaglutide

Non-arteritic anterior ischaemic optic neuropathy is a sudden loss of blood flow to the optic nerve that causes painless vision loss in one eye. A 2024 study in JAMA Ophthalmology from one Boston neuro-ophthalmology service found NAION was more common in patients prescribed semaglutide: a hazard ratio of 4.28 in type 2 diabetes and 7.64 in overweight or obesity. It was observational and drew on patients already referred to eye specialists, so the absolute rates it reported do not apply to the general population.

The EMA reviewed all the evidence, including trials, post-marketing reports and several larger cohort studies. On 6 June 2025 PRAC concluded that NAION is a very rare side effect of semaglutide (Ozempic, Rybelsus and Wegovy), affecting up to 1 in 10,000 people. The larger studies in type 2 diabetes suggested roughly a doubling of risk, about one extra case per 10,000 person-years of treatment. The advice: sudden loss of vision or rapidly worsening eyesight needs prompt medical attention, and semaglutide should be stopped if NAION is confirmed.

Aspiration and anaesthesia

Because GLP-1 drugs slow the stomach, food can still be there at the time of surgery, raising the risk of inhaling it under sedation. Early advice in 2023 was to hold weekly drugs for a week before surgery. In October 2024, guidance from five societies, including the American Society of Anesthesiologists and the American Gastroenterological Association, softened this. Most patients can continue the drug. Those at higher risk (in the dose escalation phase, on higher or weekly doses, or with nausea, vomiting, bloating or constipation) may be asked to follow a liquid diet for 24 hours before the procedure. Holding the drug is reserved for cases where the team judges the risk unacceptable. The authors call it guidance, not an evidence-based guideline. Tell any surgical or endoscopy team you are on a GLP-1 and when you last took it.

For the trials behind the benefits, see GLP-1 clinical trials. For lean tissue, see the lean mass studies.

Logging it

Most of these events are rare, and the ones that matter are timed. A log of dose, dose step, symptoms and labs on one timeline lets you and a prescriber see whether a new symptom followed a dose increase. Peptide AI keeps that timeline, and the logging is free; the free GLP-1 tracker page has the details.