The recent increase in GLP-1 (glucagon-like peptide-1) receptor agonist use not only for diabetes but also for weight loss has dominated headlines over the past few years. However, how these drugs affect one’s mental well-being is less well understood. There have been concerns about the impact GLP-1 receptor agonists have on suicidal thoughts. Some have also touted the potential benefits to mental health these drugs pose, including reduced substance use and improvements in mood. Better clarifying how GLP-1s influence mental health, particularly among those with pre-existing psychiatric illness, is critical for informing treatment decisions and understanding who the best candidates for such medications may be.
To begin to address this gap in the literature, Taipale and colleagues (2026) used the national Swedish health registers to compare the use of GLP-1 receptor agonists with non-use and with use of other second-line antidiabetic medications among individuals with a pre-existing diagnosis of depression or anxiety. They were specifically interested in understanding whether the use of this medication is associated with worsening mental health, as well as worsening of substance use disorder and self-harm. The authors used a within-individual design, comparing these interventions within the same person (i.e., how a person taking a GLP-1 compares with when that same person is not taking a GLP-1).
Methods
The authors used Swedish national health registers to gather data from 95,490 participants with a premorbid diagnosis of depression or anxiety who were prescribed non-insulin antidiabetic medication between 2009 and 2022, which were analysed in this study using a within-individual design.
The authors compared each person’s risk of worsening mental health (defined as a combination of psychiatric hospitalisations, sick leave from work exceeding 14 days for psychiatric reasons, hospitalisation due to self-harm, or death by suicide) during periods of GLP-1 use against their risk when they were not using the medication. Secondary analyses also compared different types of antidiabetic medications with one another within the same individuals. The within-individual design the authors used eliminates the effects of characteristics that don’t change over time (e.g., personality, other comorbidities).
Results
Among the 95,490 participants included in this study, the majority were female (59.7%) and had a diagnosis of anxiety (81.5%), though more than half had a diagnosis of depression (54.9%) and about a third (36.4%) had both conditions. About a quarter (23.5%) of the total cohort were prescribed GLP-1 receptor agonists, of whom 59.8% were prescribed semaglutide and 46.9% liraglutide (some people used more than one GLP-1 during follow-up).
| Outcome | Semaglutide | Liraglutide | Dulaglutide | Exenatide | Any GLP-1 receptor agonist |
|---|---|---|---|---|---|
| Worsening mental illness (main outcome) | 0.58 (0.51 to 0.65) | 0.82 (0.76 to 0.89) | 1.01 (0.85 to 1.20) | 1.01 (0.69 to 1.46) | 0.76 (0.71 to 0.82) |
| Psychiatric or self-harm hospitalisation | 0.72 (0.59 to 0.89) | 0.89 (0.76 to 1.05) | 1.09 (0.83 to 1.44) | 1.28 (0.61 to 2.68) | Not reported |
| Sick leave over 14 days for psychiatric reasons | 0.55 (0.47 to 0.64) | 0.88 (0.80 to 0.97) | 0.92 (0.73 to 1.16) | 1.14 (0.73 to 1.76) | Not reported |
| Worsening depression | 0.56 (0.44 to 0.71) | 0.74 (0.64 to 0.87) | 0.71 (0.51 to 1.01) | 1.40 (0.74 to 2.62) | 0.70 (0.62 to 0.79) |
| Worsening anxiety | 0.62 (0.52 to 0.73) | 0.91 (0.81 to 1.02) | 1.11 (0.86 to 1.44) | 1.20 (0.73 to 1.99) | 0.84 (0.76 to 0.92) |
| Worsening substance use disorder | 0.53 (0.35 to 0.80) | 1.05 (0.80 to 1.38) | 1.01 (0.60 to 1.72) | Too few events | 0.84 (0.68 to 1.04) |
| Hospital-treated self-harm | 0.51 (0.20 to 1.28) | 0.54 (0.28 to 1.07) | 0.39 (0.14 to 1.09) | Too few events | 0.56 (0.34 to 0.92) |
Use of both semaglutide and liraglutide was associated with decreased worsening of mental health problems compared with periods when the same individuals did not use GLP-1s. When semaglutide was compared with other GLP-1 Receptor agonists, semaglutide was associated with a significantly decreased risk of worsening mental health problems. Women and men did not significantly differ in their decreased risk of worsening mental health problems across all GLP-1 receptor agonists except liraglutide, for which there was only a decreased risk of worsening mental health problems in women, but not in men.
Regarding specific factors of worsening mental health problems, semaglutide was associated with a reduced risk of worsening depression, worsening anxiety, and worsening substance use, but liraglutide was only associated with a reduced risk of worsening depression. Any use of GLP-1 receptor agonists was associated with reduced risk of self-harm. Among inpatient events, semaglutide was associated with a reduced risk of psychiatric or self-harm hospitalisation. Both semaglutide and liraglutide were also associated with a lower risk of sick leave due to any psychiatric reason. When compared with other second-line antidiabetics, semaglutide was again associated with a lower risk of worsening mental health problems, whereas dapagliflozin and sitagliptin were linked to a greater risk of worsening mental health problems compared with empagliflozin.
| Medication | Adjusted hazard ratio (95% confidence interval) | Compared with empagliflozin |
|---|---|---|
| Semaglutide | 0.73 (0.62 to 0.87) | Lower risk |
| Liraglutide | 1.14 (1.00 to 1.30) | No clear difference |
| Dapagliflozin | 1.24 (1.02 to 1.52) | Higher risk |
| Sitagliptin | 1.49 (1.31 to 1.68) | Higher risk |
Conclusions
The authors concluded that, among people with established depression or anxiety who were also being treated with an antidiabetic, semaglutide and (to a lesser extent) liraglutide were associated with significantly lower risk of worsening mental health problems compared with periods of time when they were not using these medications, whereas exenatide and dulaglutide were not.
They suggest that these findings indicate differing GLP-1s may have different benefits, though they note that GLP-1 receptor agonists as a group were associated with a lower risk of self-harm (the study lacked the power to test individual drugs). The authors call for a future randomised controlled trial of GLP-1 receptor agonist use in people with diabetes and depression, anxiety, or both psychiatric conditions to improve understanding of the specific benefits GLP-1 receptor agonists may have in these populations.
Strengths and limitations
The use of a within-individual design is by far the greatest strength of this study. By having each person act as their own comparator, the study removes the effect of stable differences between people who take GLP-1s and those who don’t (such as personality or baseline illness severity). It can’t rule out factors that change over time, such as why someone starts or stops the drug, and the authors acknowledge possible residual confounding. Moreover, the very large sample size gives the study the power to detect modest associations.
One limitation of this study is the absence of clinical severity measures. The authors were unable to examine to what extent the reduced risk of worsening mental health problems seen during medication use periods is a reflection of changes in weight or symptom severity that might be expected during medication use. Additionally, there were significantly smaller samples in analyses of dulaglutide and exenatide. Therefore, the null findings for those drugs may be an issue of power rather than an absence of effect.
Another limitation of this study is that there were varying indications for the use of GLP-1 receptor agonists among the study cohort, and it is unclear whether the indication (for example, someone taking the medication for diabetes vs. for weight loss) was associated with differences in risk of worsening mental health problems. Similarly, since the majority of participants included in this study were prescribed these medications for diabetes, replicating this study with a sample of participants only taking GLP-1 receptor agonists for weight loss is critical to generalise the findings to an ever-increasing group of individuals now prescribed GLP-1 receptor agonists for this indication.
Implications for practice
This study is a necessary first step in understanding how GLP-1 receptor agonists may affect mental health among individuals with diabetes or obesity and comorbid depression or anxiety. The findings suggest that perhaps GLP-1 use may help with mental health alongside physical health. However, a randomised controlled trial is needed to better understand these preliminary findings.
For prescribers, these results also suggest that each GLP-1 receptor agonist may have different effects on mental health. The findings also suggest that semaglutide may be a sensible option for people with diabetes and a pre-existing diagnosis of depression or anxiety; the case for liraglutide is weaker, as it did no better than empagliflozin in the head-to-head comparison. However, given that these findings have not yet been replicated in a randomised controlled trial and because of individual differences, it is important to assess and monitor changes in each individual’s mental health symptoms throughout treatment and adjust accordingly.
Statement of interests
Alexandra Allam has no conflicts of interest to declare, but acknowledges AI (Claude, Anthropic) use to help with the editing of the blog text.
Editor
Edited by Dr Dafni Katsampa.
Links
Primary paper
Heidi Taipale, Mark Taylor, Markku Lähteenvuo, Ellenor Mittendorfer-Rutz, Antti Tanskanen, Jari Tiihonen. (2026) Association between GLP-1 receptor agonist use and worsening mental illness in people with depression and anxiety in Sweden: a national cohort study. Lancet Psychiatry 2026; 13: 327–35.