GLP-1 Drugs and Suicidal Thoughts: The Warning Came Off in 2026

In February 2026 the Suicidal Behavior and Ideation section was removed from the Wegovy, Zepbound and Saxenda labels.^[4][5][6] Each of those labels now contains a single mention of the word, in the table recording that the section is gone. That happened after three large studies looked for a signal and none of them found one — one of them finding the opposite. If you have read that these drugs carry a suicidality warning, that was true, and it is not true now.

Where the concern came from

In 2023 European regulators opened an investigation after reports of suicidal thoughts in people taking semaglutide.^[3] That was the right response to a safety signal: reports of that kind are exactly what post-marketing surveillance exists to catch, and the correct next step is to find out whether the drug is causing them or whether they would have occurred anyway in a large population.

There was also a plausible reason to worry. Bariatric surgery, the older standard treatment for obesity, has a small but real association with increased suicidality. A drug producing comparable weight loss raised a reasonable question about whether it might carry a comparable risk.

What the studies found

The register study is worth reading past its headline. Suicide death alone showed no significant difference. The composite of suicide death and non-fatal self-harm came out lower on GLP-1 drugs, at a hazard ratio of 0.83 (95% CI 0.70–0.97). New diagnoses of depression and anxiety were flat, at 1.01 (0.97–1.06).^[2]

And its authors did the thing good studies do, which is to say how much uncertainty is left rather than declaring the matter closed. The upper bound of their confidence interval is compatible with an absolute increase of no more than 0.16 events per 1,000 person-years.^[2] That is the honest ceiling on what the data can still hide.

What the label change actually says

Each of the three labels carries a Recent Major Changes table at the top, which records what was added or taken out and when. All three record the same thing in February 2026: the Suicidal Behavior and Ideation section, removed.^[4][5][6]

Suicidal Behavior and Ideation (5.10) ……… (Removed) 02/2026

— WEGOVY prescribing information, Recent Major Changes

A regulator removing a warning is a stronger statement than a regulator declining to add one. Warnings are cheap to keep and expensive to take off, because taking one off invites exactly the criticism a regulator least wants. It happened across three products in the same month.

What this does not settle

• Only 27 of 144 eligible trials systematically recorded these events.^[1] The meta-analysis is built from the minority of trials that were looking, which is a real limitation and its authors say so.
• The register study is mostly people with type 2 diabetes, at a mean age of 60.^[2] That is not the same population as a 38-year-old taking Wegovy for weight.
• Absence of a population signal is not a statement about you. A drug can fail to raise risk across 300,000 people and still be part of what is happening to one person.
• Continued monitoring is the authors’ own recommendation, not a formality — extended use in far larger numbers is still accumulating.^[1]

If your mood has changed since starting

Tell your prescriber, and do not talk yourself out of it because a study said the risk is not elevated. Population evidence describes populations. Rapid weight loss changes sleep, eating, hormones and how people relate to their bodies, and any of that can move mood without the drug acting on the brain directly.

It is also worth knowing that stopping abruptly is not obviously the answer either, and is not a decision to make alone. What the evidence supports is telling someone, promptly, and letting a clinician who knows your history weigh it. If you are in crisis, 988 in the US is available now, and is the right first call rather than the last one.

The randomized evidence, pooled

A 2025 meta-analysis pooled 25 randomized controlled trials and found no significant association between these drugs and suicidal behavior — a risk ratio of 0.84 (95% CI 0.54–1.32, P = 0.46).^[4]

What makes it unusually clean is the consistency. Every subgroup pointed the same way and every one had an I² of 0%, meaning no detectable variation between studies: type 2 diabetes (0.74) and obesity (1.07), adolescents (0.91) and adults (0.84), suicidal ideation (1.04), attempts (0.68) and completed suicides (1.06), and each individual drug from dulaglutide (0.46) to lixisenatide (1.25).