News · Heart & Metabolic
Stopping GLP-1 drugs tracked with more anxiety after
A Nature Metabolism study of Shanghai health records found GLP-1 drugs carried no extra psychiatric risk during treatment, but higher risk of depression and anxiety after people stopped.
Based on a peer-reviewed cohort study in Nature Metabolism
- Researchers used the Shanghai Hospital Link Database to compare people with type 2 diabetes on GLP-1 drugs against two other diabetes drug classes, publishing in Nature Metabolism.
- The comparison drugs were DPP-4 inhibitors and SGLT2 inhibitors, both used for the same condition, which makes the groups more alike than a general population comparison would.
- During treatment, GLP-1 drugs showed neutral risk against DPP-4 inhibitors and lower risk against SGLT2 inhibitors.
- After stopping, prior GLP-1 use tracked with higher risk of depressive and anxiety disorders than prior use of either comparison drug.
- The increase was modestly mediated by raised triglyceride levels, which is a partial mechanism rather than an explanation.
- The psychiatric safety picture during treatment has been inconsistent across studies; what happens after stopping was largely unexplored.
- This is electronic health record data from one city, in people with type 2 diabetes, not in people taking these drugs for weight alone.
- Observational. People who stop a drug differ from people who continue, often for reasons that also affect mood.
Almost all the argument about the psychiatric safety of GLP-1 drugs has been about what happens while people take them. A study in Nature Metabolism looked at the other end, and found the signal was not where the debate has been.
GLP-1RA discontinuation is associated with an increased risk of psychiatric events, while the treatment period itself looked unremarkable.
The question the debate skipped
Safety claims about these drugs have swung in both directions, without settling. The psychiatric safety profile of GLP-1 receptor agonists is characterized by inconsistent findings during active treatment.
The gap has been on the far side of that. Until now the profile remains largely unexplored after cessation, despite discontinuation being extremely common: people stop for cost, for side effects, or because supply runs out.
How the comparison was built
The data comes from large-scale electronic health records from the Shanghai Hospital Link Database, covering people with type 2 diabetes.
The choice of comparison group is what gives the analysis its footing. Incident depressive and anxiety disorders were compared against dipeptidyl peptidase-4 inhibitors or sodium-glucose cotransporter-2 inhibitors, two other drug classes used for the same condition.
Everyone being compared therefore has diabetes and is being treated for it. That removes a great deal of the difference between medicated and unmedicated people, which is where this kind of study usually goes wrong.
During treatment, then after
While people stayed on the drug, nothing alarming appeared. During treatment, GLP-1 drugs show a neutral risk compared with DPP4is and a lower risk compared with SGLT2is.
After stopping, the picture inverted. Prior use of GLP-1 drugs is associated with a higher risk than prior use of DPP4is or SGLT2is.
Same people, same database, same comparison drugs. The only thing that changed is which side of the stop date the follow-up sits on.
The partial mechanism
The authors offer one thread of explanation. The increase is modestly mediated by elevated triglyceride levels.
Modestly is carrying weight in that sentence. A mediation analysis in observational data describes how much of a statistical association travels through a given pathway; it does not establish that the pathway is causal. It is a lead, not a mechanism.
What the Shanghai records cannot show
The obvious alternative is reverse causation, and it is a strong one. People stop medication because of side effects, cost, weight regain or deteriorating health, and every one of those predicts low mood independently.
The setting is also specific: one city’s hospital records, in people with type 2 diabetes. Whether the same pattern holds in the much larger group taking these drugs for weight alone is not something this cohort can answer.
The authors’ conclusion is a call for attention rather than alarm: these findings highlight the need for clinical vigilance regarding anxiety following GLP-1RA discontinuation.
For anyone coming off one of these drugs, that is the useful version. Not a reason to keep taking it, and not a reason to stop, but a reason for someone to be paying attention in the weeks afterwards.
People also ask
Does stopping a GLP-1 drug cause anxiety?
This cannot show that, and the alternative explanation is strong. People stop these drugs for reasons that predict mood problems on their own: side effects, cost, weight regain, or an illness getting worse. Any of those would produce the same pattern with the arrow reversed. What the study establishes is that the period after stopping is worth watching, not that the stopping is the cause.
Why compare against other diabetes drugs?
Because it narrows the gap between the groups. Everyone in the comparison has type 2 diabetes and is on treatment for it, so the study is not comparing medicated people against the general population. That removes a large slice of confounding, though it cannot remove the reasons a clinician chose one drug over another.
What does the triglyceride finding mean?
The researchers report that the increased risk was modestly mediated by elevated triglyceride levels, meaning a portion of the statistical association runs through that pathway. Modestly is the operative word: it is a partial account, in observational data, and mediation analysis assumes a causal structure it cannot itself verify.
Does this apply to people taking these drugs for weight loss?
Not directly. The cohort is people with type 2 diabetes, and the comparison drugs are diabetes treatments. People using GLP-1 drugs purely for weight management differ in age, health and reasons for stopping. Whether the same pattern appears in that group is an open question this study does not answer.
Should someone stop taking their medication because of this?
No, and that is not what the finding suggests. This is general information rather than advice. If anything the practical reading runs the other way: it points at the period after stopping as a time to be attentive to mood. Any decision about starting or stopping a prescribed medicine belongs with the prescribing clinician.
References
- Zhang T, He P, Ji Y, et al. Association of GLP-1RA discontinuation and risk of depressive and anxiety disorders in people with type 2 diabetes: a cohort study. Nature Metabolism (2026).
- National Institute of Diabetes and Digestive and Kidney Diseases. Insulin, Medicines, and Other Diabetes Treatments.