What this is
- This research investigates psychiatric adverse events (AEs) linked to glucagon-like peptide 1 (GLP-1) analogues using data from US, Canada, and Australia.
- The study identifies significant associations between GLP-1 analogues and various psychiatric AEs, particularly with semaglutide and liraglutide.
- It emphasizes the need for cautious prescribing and further research to understand these associations, especially in patients with pre-existing psychiatric conditions.
Essence
- GLP-1 analogues, especially semaglutide and liraglutide, are associated with significant psychiatric adverse events, including depression and suicidal ideation. The study calls for further investigation into these associations.
Key takeaways
- Semaglutide is linked to depressive symptoms ( = 6.24) and suicidal ideation ( = 2.58) in the US database. These findings indicate a strong association between semaglutide use and serious psychiatric AEs.
- Liraglutide shows a significant association with depression (CVAROD, = 1.68). This suggests that patients using liraglutide may also experience heightened psychiatric risks.
- Dulaglutide is associated with eating disorders ( = 1.47) and insomnia ( = 2.93), indicating that this medication may also contribute to specific psychiatric AEs.
Caveats
- The study relies on self-reported AEs, which may lead to bias in reporting. This limitation affects the reliability of the observed associations.
- Lack of detailed patient characteristics, such as medical history, complicates the ability to determine causal relationships between medication use and AEs.
- The total number of patients prescribed each medication was unknown, limiting the ability to assess the true incidence of AEs associated with GLP-1 analogues.
Definitions
- Reporting Odds Ratio (ROR): A statistical measure used to determine the strength of association between medication use and adverse events.
AI simplified
Impact statements
By demonstrating potential associations between several psychiatric adverse events and glucagon-like peptide 1 analogues, this research supports the need for further investigations to establish these associations. It is important to consider the patient's psychiatric health and history before prescribing to ensure the prescribing of glucagon-like peptide 1 analogues is appropriate. This study compares the risks associated with each member of the class, highlighting the potential differences in psychiatric adverse event profiles of various glucagon-like peptide analogues, and suggests that further prospective studies are warranted for confirmation.
Introduction
The class of drugs known as glucagon-like peptide 1 receptor agonists (GLP-1 RAs), also called GLP-1 analogues, are synthetic analogues of GLP-1 which act by stimulating glucose-dependent insulin release from the pancreas and slowing gastric emptying [1]. The primary aim of these medications is to increase glucose-dependent insulin release and to inhibit inappropriate glucagon secretion to achieve ideal diabetic control [1]. Alongside glycaemic control, GLP-1 analogues have also been shown to have a favorable cardiovascular profile and contribute to weight loss [1, 2]. As a result, prescribing of these medications has increased exponentially in recent years. In the US alone, monitoring has found that GLP-1 analogue use in people with type 2 diabetes and atherosclerotic cardiovascular disease nearly doubled, rising from 5.2% in 2018 to 9.9% in 2022 [3]. Moreover, posts on social media have triggered a significant 'off-label' demand for GLP-1 analogues to achieve weight loss, further increasing the number of GLP-1 analogue users [4].
GLP-1 analogues have had extensive investigations into their safety and efficacy with several diverse adverse events (AEs) being associated with their use. Gastrointestinal AEs are by far the most common, ranging from nausea and diarrhoea (in up to 50% of users) to abdominal pain and constipation (in up to 10% of users) [5, 6]. Additionally, there have been reports of rare AEs such as pancreatitis and cancer that were thought to be associated with GLP-1 analogues [6, 7]. However, the lack of investigation into potential psychiatric AEs raises a concern for those with pre-existing psychiatric conditions that could benefit from GLP-1 analogue use. Even though some recent studies have explored the psychiatric effects of these medications, these studies have been contradictory and inconclusive [8 –10]. Two studies have reported the beneficial effects of GLP-1 analogues on depression and anxiety in a corticosterone-induced depression mice model or in patients newly diagnosed with type 2 diabetes [8, 9]. Another study by Wang et al. has also found that GLP-1 analogues, specifically semaglutide, are linked to a significantly lower risk of suicidal behaviors compared to non-GLP-1 analogues for either treatment of obesity (e.g. naltrexone, bupropion, etc.) or diabetes (e.g. metformin, insulin etc.) [10]. Conversely, one study found an association between their use and increased psychiatric effects, including nervousness, insomnia, and eating disorders [11]. Additionally, a study performed by Ruggiero et al. [12] found that semaglutide and liraglutide were associated with a two-to-four-fold increase in the reporting probabilities of suicidal events compared to exenatide and dulaglutide. Similarly, a recent study by Guirguis et al. [13] showed a potential relationship between liraglutide, semaglutide, and tirzepatide and suicidal thoughts and behaviors, further highlighting the need for further research.
Analysing 'real-world' data may lead to the emergence of previously unreported AEs, sparking interest in further investigations for new user groups. The US Food and Drug Administration Adverse Events Reporting System (FAERS), Health Canada's Vigilance Adverse Reaction Online Database (CVAROD) and Australian Therapeutic Goods Administration's Database of Adverse Event Notifications (DAEN) are government databases to which healthcare professionals can report suspected post-market AEs with the aim of helping agencies and researchers identify safety concerns in pharmaceutical products [14 –16]. Many pharmacovigilance studies have previously used data available in these government databases to report AE trends in a 'real-world' setting [17, 18].
A recent pharmacovigilance study examining psychiatric AEs associated with semaglutide, liraglutide and tirzepatide in the European Medicines Agency EudraVigilance database [19] reported depression, anxiety and suicidal ideation as the most common psychiatric AEs, with 20 out of 372 reports had a fatal or life-threatening outcome. However, as a disproportionality analysis was not completed, it was not possible to determine if there is an association between the use of these GLP-1 analogues and the psychiatric AEs examined. Further, this study only explored psychiatric AEs in three GLP-1 analogues, rather than all six agents within the class. Another pharmacovigilance study [11] investigated psychiatric AEs associated with GLP-1 analogues using the FAERS up until Q1 2023 found associations between these drugs and eight categories of psychiatric AEs. Given the rapidly increase in GLP-1 analogue usage globally, examination of psychiatric AEs in multiple large spontaneous adverse events reporting databases is required to gain a current understanding of the relationship between the use of GLP-1 analogues and psychiatric AEs.
Aim
Combining and comparing AE reports from multiple databases enriches the analyses by increasing the sample size, breadth of data representation, and statistical power to detect associations. Hence, this study aimed to investigate the prevalence of psychiatric AEs associated with currently available GLP-1 analogues by analysing publicly available national datasets from the US (FAERS), Canada (CVAROD) and Australia (DAEN).
Ethics approval
Ethical approval was not required for our study as it did not involve human participants, animals or the collection of personal data. Exemption was granted by the RMIT University ethics committee.
Method
Data sources
AE reports for all approved GLP-1 analogues (exenatide, semaglutide, liraglutide, lixisenatide, dulaglutide and tirzepatide) were downloaded from three online databases: FAERS, CVAROD and DAEN. The study period for each drug consisted of data from the launch date of each drug in the respective countries to the latest available data at the time of the search (Supplementary Table ). 1
Search strategies
A search of all psychiatric AEs was conducted for the GLP-1 analogues in the respective databases. To expedite and ensure uniformity across all databases, the Medical Dictionary for Regulatory Activities (MedDRA) classification of psychiatric events was used since all three databases had used the MedDRA classification system to categorise the reported AEs. Any events that did not fit into this classification were excluded. Similar AEs were clustered together in categories for ease of understanding (e.g. insomnia include insomnia, initial insomnia, middle insomnia and terminal insomnia).
The number of psychiatric events were recorded alongside the number of patients experiencing these events, their age and sex. Since concomitant medications and patient medical history were not fully known or recorded, these patient factors were not included in the analyses.
Data analysis
A disproportionality analysis model was employed to identify associations between GLP-1 analogue use, and the occurrence of a specific AE compared to all other drugs (reported in that database). To ensure that associations are not driven by isolated findings from a single medication, all AEs from the three databases were reviewed and reported if they met all of the following criteria: At least 10 reports for a single medication in a single database, and Had been reported for at least two medications in at least two databases.
The minimum total number of reports was typically defined as 3 for ROR analysis [11]. As this is a comprehensive analysis involving all GLP-1 analogues utilising data from three national databases, we have established this inclusion criteria to minimise the chance of false positive signals caused by small reporting numbers, and to provide a clearer overview of adverse events with potential clinical relevance.
Reporting Odds Ratio (ROR) with 95% confidence intervals (CI) were calculated for each AE of interest, for each medication across the three databases [20]. An ROR greater than 1 indicates a greater correlation between the drug and AE. The Reporting of A Disproportionality analysis for drug Safety signal detection using individual case safety reports in Pharmacovigilance (READUS-PV) checklist was used to report this disproportionality analysis study (Supplementary Data File 2) [21, 22].
The formula for calculating ROR is: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\frac{\left(\frac{a}{c}\right)}{\left(\frac{b}{d}\right)}$$\end{document} a c b d
where a is the number of reported cases of a symptom associated with the medication of interest; b is the number of other symptoms reported for that same medication; c is the number of reported cases of the same symptom associated with all other medications, and d is the number of all other symptoms associated with all other medications. A resulting odds ratio of greater than 1 is considered to suggest that the medication in question is more likely to be associated with reports of that symptom than are other medications [20].
The 95% CI is calculated using the formula: \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${e}^{\text{ln}\left(ROR\right)\pm 1.96\sqrt{\frac{1}{a}+\frac{1}{b}+\frac{1}{c}+\frac{1}{d}}}$$\end{document} e ln R O R ± 1.96 1 a 1 b 1 c 1 d + + +
This formula gives us two values representing the upper and lower confidence interval. If the range between these values does not include 1 then our result is considered significant.
The US database is the largest dataset in this study, providing a larger sample size for calculating RORs leading to more robust results. In contrast, the Australian and Canadian datasets are significantly smaller, which limited the power of their individual findings. A small number of reported cases could return very high RORs but with a very wide range in the CI, signifying less reliable results. Because of this, the US RORs were used to guide targeted searches in the smaller datasets.
While findings in smaller datasets should be interpreted cautiously, alignment with the larger US dataset increased confidence in the observed associations. Any cross-dataset agreement suggests the findings may be applicable across different population groups.
Results
Basic information on AE reports
A total of 16,678 psychiatric reports were identified for the GLP-1 analogues of interest across all three databases: exenatide (n = 5492, 32.9%), semaglutide (n = 4067,24.4%), dulaglutide (n = 3024, 18.1%), tirzepatide (n = 1447, 8.7%), liraglutide (n = 2552,15.3%) and lixisenatide (n = 96, 0.6%) (Supplementary Table ). The three databases recorded the following number of reports each: FAERS (n = 16,090, 96.5%), CVAROD (n = 363, 2.2%) and DAEN (n = 225,1.3%) (Supplementary Table ). Majority of the reports were in females (n = 7990, 64.2%) and the median age across the three databases was 60 years [interquartile range (IQR) = 54–68] (Supplementary Tablesand). 2 2 3 4
Furthermore, among the AEs investigated, the most frequently reported in each database were explored further. In FAERS, depressive disorders (n = 1,956, 14%), insomnia (n = 1,861, 13.2%) and anxiety (n = 1,851, 13.2%) were the top reported AEs. In CVAROD, depressive disorders (n = 65, 19.5%), suicidal ideation (n = 45, 13.5%) and anxiety (n = 41, 12.3%) were the most common. Meanwhile, in DAEN, suicidal ideation (n = 50, 25.1%), depressive disorders (n = 30, 15.1%) and anxiety (n = 24, 12.1%) were the highest reported (Table 1). Most psychiatric AEs were recorded from patients taking exenatide (n = 4,777, 32.9%), followed by semaglutide (n = 3,490, 24.1%), dulaglutide (n = 2,667, 18.4%), liraglutide (n = 2,215, 15.3%), tirzepatide (n = 1,282, 8.8%) and lixisenatide (n = 79, 0.5%) (Table 1).
| Exenatide | Semaglutide | Dulaglutide | Tirzepatide(5) | Liraglutide | Lixisenatide(5) | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| FAERS | DAEN | CVAROD | FAERS | DAEN | CVAROD | FAERS | DAEN | CVAROD | FAERS | DAEN | FAERS | DAEN | CVAROD | FAERS | CVAROD | |
| Agitation and restlessness | 394 | 1 | – | 151 | 3 | 11 | 152 | 2 | – | 70 | 1 | 141 | 2 | 6 | 6 | – |
| Agitation | 90 | – | – | 28 | 1 | 1 | 33 | 1 | – | 15 | – | 33 | 2 | 1 | 2 | – |
| Anger | 48 | 1 | – | 22 | – | 3 | 22 | 1 | – | 12 | – | 21 | – | 1 | – | – |
| Irritability | 217 | – | – | 78 | 2 | 7 | 82 | – | – | 32 | – | 61 | – | 3 | 3 | – |
| Restlessness | 39 | – | – | 23 | – | – | 15 | – | – | 11 | 1 | 26 | – | 1 | 1 | – |
| Mood disorders | 2549 | 6 | 2 | 1554 | 37 | 76 | 1250 | 9 | – | 577 | 1 | 955 | 24 | 69 | 37 | 1 |
| Mood altered | 44 | – | – | 35 | 1 | 5 | 29 | – | – | 16 | – | 26 | 1 | 4 | – | – |
| Mood Swings | 39 | – | – | 26 | 1 | 1 | 24 | 1 | – | 8 | – | 36 | – | 2 | – | – |
| Depressive disorders(1) | 520 | 4 | – | 609 | 14 | 34 | 290 | 2 | – | 174 | – | 354 | 10 | 31 | 9 | – |
| Apathy | 13 | – | – | 26 | 1 | 2 | 12 | – | – | 5 | – | 14 | 1 | 3 | – | – |
| Emotional disorder | 23 | – | – | 18 | 2 | – | 17 | 1 | – | 6 | – | 13 | – | 1 | – | – |
| Anxiety | 574 | 1 | 1 | 434 | 11 | 19 | 327 | 2 | – | 217 | 1 | 291 | 9 | 21 | 8 | – |
| Nervousness | 695 | – | – | 113 | – | 3 | 163 | 1 | – | 33 | – | 90 | – | 2 | 10 | – |
| Emotional distress | 41 | – | – | 27 | 1 | – | 13 | – | – | 9 | – | 10 | – | 2 | 1 | 1 |
| Panic attack | 54 | – | – | 77 | 3 | 2 | 37 | – | – | 53 | – | 33 | 2 | 1 | 1 | – |
| Post-traumatic stress disorder | 8 | – | – | 6 | 1 | – | 17 | 1 | – | 2 | – | 8 | 1 | – | – | – |
| Stress | 462 | 1 | 1 | 157 | 2 | 8 | 282 | – | – | 45 | – | 70 | – | 2 | 6 | – |
| Fear | 76 | – | – | 26 | – | 2 | 39 | 1 | – | 9 | – | 10 | – | – | 2 | – |
| Confusion and cognitive changes | 744 | 3 | – | 262 | 11 | 13 | 378 | 6 | 1 | 89 | – | 214 | 5 | 8 | 11 | – |
| Confusional state | 424 | 2 | – | 141 | 4 | 6 | 194 | 3 | – | 37 | – | 115 | – | 4 | 6 | – |
| Disorientation | 183 | – | – | 33 | – | 1 | 34 | – | – | 14 | – | 32 | – | 2 | – | – |
| Mental disorder | 45 | 1 | – | 41 | 5 | 2 | 71 | 2 | – | 19 | – | 25 | 4 | 1 | 2 | – |
| Delirium | 18 | – | – | 12 | 1 | 1 | 10 | 1 | – | 4 | – | 15 | – | 1 | 1 | – |
| Thinking abnormal | 48 | – | – | 16 | – | 1 | 53 | – | – | 9 | – | 18 | 1 | – | 1 | – |
| Abnormal behaviours | 26 | – | – | 19 | 1 | 2 | 16 | – | 1 | 6 | – | 9 | – | – | 1 | – |
| Eating disorder | 83 | 1 | – | 44 | – | 5 | 172 | 1 | – | 36 | – | 33 | – | 2 | 2 | – |
| Psychotic disorders | 68 | 3 | – | 102 | 4 | 7 | 53 | – | – | 20 | – | 30 | 2 | 2 | 1 | – |
| Hallucinations(2) | 53 | 2 | – | 51 | – | 5 | 34 | – | – | 10 | – | 19 | 1 | 2 | 1 | – |
| Paranoia | 11 | – | – | 9 | 2 | – | 11 | – | – | 3 | – | 5 | 1 | – | – | – |
| Psychotic disorder | 4 | 1 | – | 42 | 2 | 2 | 8 | – | – | 7 | – | 6 | – | – | – | – |
| Sleep-related disorders | 820 | 5 | – | 684 | 4 | 45 | 565 | 3 | 1 | 408 | – | 456 | 9 | 32 | 20 | – |
| Sleep disorders(3) | 105 | – | – | 152 | 2 | 15 | 216 | 1 | 1 | 115 | – | 105 | 2 | 8 | 6 | – |
| Insomnia(4) | 608 | 3 | – | 423 | 1 | 21 | 284 | 1 | – | 241 | – | 293 | 5 | 18 | 12 | – |
| Nightmare | 32 | 1 | – | 49 | – | 4 | 16 | 1 | – | 12 | – | 22 | 1 | – | 1 | – |
| Abnormal dreams | 25 | – | – | 38 | 1 | 4 | 13 | – | – | 18 | – | 14 | – | 3 | – | – |
| Poor quality sleep | 50 | 1 | – | 22 | – | 1 | 36 | – | – | 22 | – | 22 | 1 | 3 | 1 | – |
| Suicidal thoughts and behaviours | 95 | 3 | – | 422 | 25 | 30 | 73 | 1 | – | 80 | – | 176 | 27 | 22 | 1 | – |
| Completed suicide | 5 | – | – | 28 | – | 2 | 4 | – | – | 2 | – | 27 | – | 1 | – | – |
| Depression suicidal | 7 | – | – | 26 | – | 1 | 2 | – | – | 4 | – | 4 | 3 | – | – | – |
| Suicidal ideation | 65 | 3 | – | 325 | 23 | 26 | 47 | 1 | – | 72 | – | 98 | 23 | 19 | – | – |
| Suicide attempt | 18 | – | – | 43 | 2 | 1 | 20 | – | – | 2 | – | 47 | 1 | 2 | 1 | – |
Reporting odds ratio results
These results suggest a statistically significant association between each medication and various psychological symptoms, with some symptoms (e.g., depressive symptoms for semaglutide and sleep disorders for tirzepatide) showing particularly strong associations.
| Medication | Symptom | ROR | Lower CI | Upper CI |
|---|---|---|---|---|
| Exenatide | Disorientation | 1.32 | 1.14 | 1.52 |
| Nervousness | 3.75 | 3.48 | 4.05 | |
| Stress | 1.9 | 1.73 | 2.08 | |
| Semaglutide | Depressed mood | 1.48 | 1.25 | 1.76 |
| Depression | 1.35 | 1.22 | 1.48 | |
| Depressive symptoms | 6.24 | 4.49 | 8.69 | |
| Nervousness | 1.59 | 1.32 | 1.91 | |
| Panic attack | 1.46 | 1.16 | 1.82 | |
| Stress | 1.24 | 1.06 | 1.45 | |
| Depression suicidal | 4.35 | 2.95 | 6.4 | |
| Suicidal ideation | 2.58 | 2.31 | 2.88 | |
| Dulaglutide | Eating disorder | 1.47 | 1.26 | 1.71 |
| Sleep disorder due to general medical condition, insomnia type | 2.93 | 2.35 | 3.66 | |
| Tirzepatide | Eating disorder | 1.58 | 1.14 | 2.2 |
| Panic attack | 1.79 | 1.37 | 2.35 | |
| Sleep disorder due to general medical condition, insomnia type | 6.13 | 4.73 | 7.95 |
| Medication | Symptom | ROR | Lower CI | Upper CI |
|---|---|---|---|---|
| Semaglutide | Depressed mood | 4.97 | 1.85 | 13.38 |
| Depression | 6.75 | 3.59 | 12.69 | |
| Panic attack | 3.56 | 1.14 | 11.12 | |
| Stress | 4.15 | 1.03 | 16.78 | |
| Dulaglutide | Eating disorder | 17.66 | 2.45 | 127.37 |
| Sleep disorder due to general medical condition, insomnia type | 259.13 | 31.07 | 2161.43 |
| Symptom | Medication | ROR | Lower CI | Upper CI | Database |
|---|---|---|---|---|---|
| Depressed mood | Semaglutide | 1.48 | 1.25 | 1.76 | FAERS |
| Semaglutide | 4.97 | 1.85 | 13.38 | DAEN | |
| Depression | Semaglutide | 1.35 | 1.22 | 1.48 | FAERS |
| Semaglutide | 6.75 | 3.59 | 12.69 | DAEN | |
| Liraglutide | 2.51 | 1.12 | 5.61 | DAEN | |
| Liraglutide | 1.68 | 1.12 | 2.51 | CVAROD | |
| Panic attack | Semaglutide | 1.46 | 1.16 | 1.82 | FAERS |
| Semaglutide | 3.56 | 1.14 | 11.12 | DAEN | |
| Stress | Semaglutide | 1.24 | 1.06 | 1.45 | FAERS |
| Semaglutide | 4.15 | 1.03 | 16.78 | DAEN | |
| Suicidal ideation | Semaglutide | 2.58 | 2.31 | 2.88 | FAERS |
| Semaglutide | 4.38 | 2.97 | 6.47 | CVAROD | |
| Eating disorder | Dulaglutide | 1.47 | 1.26 | 1.71 | FAERS |
| Dulaglutide | 17.66 | 2.45 | 127.37 | DAEN | |
| Sleep disorder due to general medical condition, insomnia type | Dulaglutide | 2.93 | 2.35 | 3.66 | FAERS |
| Dulaglutide | 259.13 | 31.07 | 2161.43 | DAEN |
Discussion
GLP-1 analogues have become increasingly popular in recent decades, primarily for their effectiveness in improving glycaemic control, promoting weight loss, and providing cardiovascular benefits [1 –3]. While numerous studies have investigated the potential psychiatric AEs associated with GLP-1 analogues, their findings remain contradictory [8 –13]. Our study was the first pharmacovigilance investigation that incorporated data from three different global databases to examine the potential psychiatric risks associated with GLP-1RAs. Using the relevant search dates (Supplementary Table 1), we identified a total of 16,678 psychiatric AE reports across the three databases for the GLP-1 analogues of interest (Supplementary Table 2). Notably, the number of psychiatric AE reports from FAERS alone (16,090) has doubled when compared to a previous pharmacovigilance study (8,240) which examined psychiatric AEs for all GLP-1 analogues in FAERS from Q1 2004—Q1 2023 [11], highlighting a significant increase in psychiatric AEs associated with GLP-1 analogues in the last 2 years.
More than 57% of reports occurred in patients aged 18 or older (42% unspecified) and about 64% were in females, highlighting the possible more frequent use of GLP-1 analogues among adult female populations. A previous study using the EudraVigilance database found similar results, showing that approximately 50% of reports came from the 18–64 year age group and 65% were female [19].
We observed several significant associations between certain GLP-1 analogues and specific psychiatric AEs. Treatment with semaglutide was highly associated with the occurrence of depressed mood, panic attack, and stress in both the FAERS and DAEN databases, and it was also notably linked to suicidal ideation in both the FAERS and CVAROD databases. Treatment with dulaglutide was highly associated with the occurrence of eating disorders and sleep disorders due to general medical condition, insomnia type in both the FAERS and DAEN databases while treatment with liraglutide demonstrated a strong relationship with depression in both the DAEN and CVAROD databases.
These findings were consistent with those of Chen et al. [11], who also reported significant associations between the overall class effect of GLP-1 analogues and symptoms such as stress, eating, and sleep disorder due to general medical condition-insomnia type. However, no positive associations were reported in their study between depression, depressed mood, panic attack, and suicidal ideation and GLP-1 analogues. Notably, a large cohort study by Kornelius et al. [23] found that GLP-1 analogues, particularly liraglutide and semaglutide, are associated with a 195% increased risk of depression and a 106% higher risk for suicidal ideation. Their findings, along with those of the present study, were also aligned with two other pharmacovigilance studies, which identified potential associations between GLP-1 analogues, particularly semaglutide, and an increased risk of suicidal thoughts and behaviors due to higher reporting probabilities [12, 13]. Conversely, a retrospective cohort study [10] reported a lower risk of incidence and recurrence of suicidal ideation in patients prescribed with semaglutide. Furthermore, a clinical trial involving 3,377 participants and a cohort study with 124,517 participants found that GLP-1 analogues were not associated with either suicidal ideation, depression, nor suicidal death [24, 25]. The observed differences across different studies suggest uncertainty regarding the psychiatric effects of GLP-1 analogues, indicating a need for further investigation.
The prevalence of psychiatric AEs with GLP-1 analogue use may be due to pre-existing psychiatric disorders in patients [11]. The databases used in this study did not contain information on any pre-existing psychiatric conditions in patients reporting AEs. While the association between GLP-1 analogue use and psychiatric AEs is still not clear, given the potentially serious consequences of psychiatric AEs, additional attention in drug selection and closer monitoring of AEs in patients is warranted. As suggested in other studies, further pharmacovigilance and clinical studies are required to establish the causality link and identify the mechanism between psychiatric AEs and the use of GLP-1 analogues [10 –13, 25].
Strengths
The use of three different databases based in three different countries provided a larger dataset than other studies to date. This increases statistical power allowing for more reliable identification of AEs that might not be apparent in country-specific studies with smaller samples. The countries included in this study have diverse populations which may influence responses to the drugs. This could offer insights into whether AE trends are consistent across different demographics, potentially highlighting safety concerns.
Additionally, the use of ROR to indicate the degree of association between drugs and AEs allows for a more straightforward comparison between each database and drug. Investigating a range of medications in a class allows for the identification of any potential differences in GLP-1 analogues and their AE profiles. This may aid clinicians in tailoring prescribing decisions in patients with pre-existing psychiatric conditions.
Limitations
This study had limitations that must be considered when interpreting the results. The study relied on self-reported AEs from patients and prescribers, and this may be subject to bias leading to over- or under-reporting. Moreover, these reports were based on the reporter's opinion, making it difficult to determine the specific cause of the event.
Another limitation was the lack of patient characteristics such as medical history, comorbidities, family history and lifestyle factors. This made it difficult to distinguish between a medicine-induced reaction and an event related to a patient's ongoing health issues. Hence, it is difficult to determine a causal relationship between medication use and AEs. Note there are currently no clinically significant drug-drug interactions with any GLP-1 analogues therefore this was not assessed [26].
The total number of patients prescribed each medication in each country was unknown, so it was not possible to determine the true incidence of the AEs. In this case the ROR is the best statistical tool available to the investigators. The newer drugs have fewer reports since they have been on the market for a shorter period which made it difficult to compare these medicines to others that had been available for longer.
Hence, whilst these reports provide signals for potential drug risks, they should be interpreted with caution and be supplemented with clinical studies to confirm associations.
Conclusion
Our study is the first pharmacovigilance study that involved multiple databases from the US, Canada and Australia in investigating the psychiatric AEs associated with GLP-1 analogues. Several significant associations between certain GLP-1 analogues and specific psychiatric AEs were observed. Although several limitations exist, this preliminary study provides valuable insights for further research into the psychiatric AEs associated with GLP-1 analogues and highlights the significance of ongoing monitoring and evaluations of these medications, especially considering the potentially serious consequences of psychiatric AEs. Healthcare professionals are encouraged to approach prescribing of these medications with caution, after weighing their potential risks and benefits.
Supplementary Information
Below is the link to the electronic supplementary material. Supplementary file1 (DOCX 163 KB)