GLP-1 Drugs and Depression: What the FDA Warning and the Studies Show

Zepbound's FDA label specifically directs monitoring for new or worsening depression, suicidal thoughts or behavior, or unusual mood changes — a warning covered briefly in GLP-1 side effects. It's a serious enough concern to warrant its own sourced look at where that warning came from and what the strongest available evidence since then has actually found.

Where the concern started

Regulatory monitoring for a drug side effect usually starts with signal detection — patterns in spontaneous adverse-event reports, not proof of causation. A 2024 pharmacovigilance analysis of the EudraVigilance database, which collects individual case safety reports submitted across the EU, examined psychiatric adverse events reported alongside semaglutide, liraglutide and tirzepatide use¹. This is exactly the kind of data that raises a regulatory flag: real reports exist, submitted by clinicians and patients, and they're worth taking seriously enough to investigate formally. What this kind of database cannot do is establish whether the drug caused the event, since there's no comparison group of similar patients who weren't taking the drug — a limitation the researchers running formal cohort studies were specifically trying to address.

The largest cohort study to test it directly

In 2024, researchers used nationwide prescription and cause-of-death registries from Sweden and Denmark to directly compare suicide risk between new users of GLP-1 receptor agonists and new users of a different diabetes drug class (SGLT2 inhibitors) as an active comparator — a design built specifically to control for the fact that people starting either drug class share similar underlying health conditions². The study followed 124,517 GLP-1 users and 174,036 SGLT2 inhibitor users for a mean of 2.5 years. The results were reassuring: 77 suicide deaths occurred among GLP-1 users versus 71 among SGLT2 inhibitor users — a hazard ratio of 1.25, with a confidence interval (0.83–1.88) that included no effect. For the composite of suicide death and nonfatal self-harm, the hazard ratio was actually lower on GLP-1 drugs, at 0.83. For incident depression and anxiety diagnoses specifically, the hazard ratio was 1.01 — essentially no difference at all. The researchers' own conclusion: this cohort study does not show an association between GLP-1 receptor agonist use and increased risk of suicide death, self-harm, or new depression or anxiety diagnoses.

Holding both facts at once

Both things are true simultaneously, and neither cancels the other out. The pharmacovigilance signal was real enough that regulators added label language directing clinicians to monitor for mood changes — that's the system working as designed, erring toward caution on a serious possible harm. The largest rigorous cohort study built specifically to test that signal, in a population of over 124,000 GLP-1 users, did not find an elevated risk. That is the current state of the evidence, not a final word — cohort studies can miss effects that are rare, delayed, or concentrated in a subgroup the study wasn't powered to detect, which is exactly why the monitoring requirement stays on the label regardless of a reassuring cohort result.

What this means in practice

The label's instruction stands on its own regardless of how the population-level statistics read: monitor for new or worsening depression, suicidal thoughts, or unusual mood changes, and contact your provider if any appear — that guidance doesn't require the cohort evidence to point one way or the other to be worth following for an individual patient. If you have a personal or family history of depression or a mood disorder, that's worth raising with your prescriber before starting treatment, not after a symptom shows up. The rest of the FDA-labeled side effects and boxed warnings for this drug class are in GLP-1 side effects, and if a change in appetite or eating pattern is part of what's prompting the question, GLP-1 microdosing and GLP-1 weight regain after stopping cover two other places patients report changes the label doesn't fully capture. The rest of the Desk's sourced evidence is collected in the research library.