The Evidence Base
May 19, 2025
In large clinical trials, glucagon-like peptide-1 (GLP-1) receptor agonists have demonstrated significant effectiveness in improving glycemic control and reducing cardiovascular risk. However, recent concerns have arisen following reports suggesting a potential association between GLP-1 receptor agonists and increased risk of suicidal ideation (SI) and self-harm. While several observational studies have not supported these claims, the possibility of a safety signal remains under scrutiny.
A new study presented at ISPOR 2025 explores this issue further, assessing whether the prevalence of SI increases after initiating GLP-1 therapy and whether this varies by treatment indication. In this interview, we hear from the study’s presenting author, Jessica Probst (Principal, Real World Evidence, OM1), about the findings featured in the poster “Suicidal Ideation in Patients Treated With Glucagon-like Peptide-1 Receptor (GLP-1) Agonists: A Retrospective Real-World Analysis.”
Jessica, thanks for speaking with us about your research presented at ISPOR 2025. What prompted OM1 to investigate the association between GLP-1 receptor agonists and suicidal ideation using real-world data?
This study was motivated by discordant warning labels for GLP-1 agonists across indications. The labels for therapies indicated for type 2 diabetes contain a warning for pancreatitis and acute kidney injury, while therapies indicated for weight loss share those same warnings plus an additional warning for suicidal behavior and ideation. GLP-1s represent a breakthrough treatment option for patients with obesity who have long faced a bias based on their weight from the healthcare system. This work seeks to support a growing body of research exploring whether the additional warning for SI for GLP-1’s indicated for weight loss may be an unnecessary obstacle to treatment and potential threat to equitable access to healthcare for a long-stigmatized population.
In this analysis, we sought to understand whether there were differences in the prevalence of SI in patients before and after treatment with a GLP-1. As a sub-analysis, we also looked for differences in SI between patients treated for type 2 diabetes separately from those treated for weight loss. We analyzed whether patients reporting ideation after treatment initiation had a history of prior SI before their GLP-1 use. This is particularly relevant for an analysis focused on package labeling, which may influence prescribing patterns or induce patient or provider hesitancy regarding use of GLP-1s for weight loss in patients with certain risk factors such as a history of suicidality.
Given the background reports of suicidal ideation and existing research in this area, what specific evidence gaps did your study aim to address?
The FDA’s evaluation of SI in patients on GLP-1 therapies was based on data from the FDA Adverse Event Reporting System (FAERS), which requires active reporting of adverse events to the FDA. The observational nature of the real-world data (RWD) used in this study eliminates the bias that results from the need for events to have been actively reported to FAERS.
While other observational studies leveraging RWD have also explored the safety profile of these therapies, these studies rely heavily on claims and structured data from electronic medical records (EMRs), which also have limitations. These approaches result in binary categorization of patients into distinct groups: 1) those with documented SI; and 2) those without documented ideation. In this second group, there is no ability to differentiate patients who indicated they did not have ideation from patients who did not indicate whether they had ideation.
This analysis leveraged the OM1 Real-World Data Cloud™ (RWDC), which contains healthcare data on over 350 million patients in the US, including EMR data on nearly 90 million patients. Attestations of SI have previously been extracted using natural medical language processing from unstructured EMR notes, supplementing evidence of SI captured via medical claims and in structured EMR fields. The extraction also distinguished between affirmation and negation of SI, enabling us to fill in the gaps produced by other EMR and claims-based approaches.
Can you tell us more about the RWDC, the robustness of the data within it, and what advantages it offers over other data sources?
The RWDC is a continuously updated, deterministically linked, multi-source dataset. The RWDC’s medical and pharmacy claims contain billing and coding history on inpatient and outpatient encounters from acute care facilities, ambulatory surgery centers, and clinics. The combined EMR and claims dataset provides further insights into the complete patient journey with the addition of laboratory data, vital signs, problem lists, and other clinical details unavailable in claims data alone.
Data within the RWDC dates back to 2013. Patients with a wide age and geographic distribution are represented including patients from all 50 states and territories. Since data are updated continuously, the RWDC is uniquely suited to longitudinally monitoring near real-time disease and treatment trends with key clinical insights that require thoughtfully curated information from EMR data sources.
How did you identify the patient cohort included in the study, and what inclusion or exclusion criteria were applied?
Patients in the study were required to have evidence of initiation a GLP-1 therapy for either type 2 diabetes or obesity, to have no record of previous prescriptions for other second-line type 2 diabetes or weight loss medications, and to have an assessment of SI in the year before and after GLP-1 treatment initiation.
What were the study’s key findings, and did any outcomes stand out as particularly noteworthy?
The study found there was no significant change in prevalence of SI after initiation of a GLP-1 medication, and that most patients who affirmed ideation after treatment initiation also affirmed ideation in the year prior to treatment start. This is important for assessing whether any potential change in ideation prevalence during follow-up would represent a new event for patients with no prior history or potential persistence or recurrence in patients with prior history of SI.
We also found the prevalence of SI after GLP-1 initiation did not significantly differ between patients with type 2 diabetes compared with patients treated for weight loss, despite differing package label warnings. The results indicated that, across indications, a similar proportion of patients treated with a GLP-1 had prior history of SI in the year before initiation.
What are the potential implications of your research for clinicians and regulators when assessing psychiatric safety signals associated with GLP-1 therapies? And from the patient’s perspective, how might these findings help inform discussions around the risks and benefits of starting or continuing treatment?
While the package labels for GLP-1 use for weight loss carry an additional warning for SI, this research did not find increased prevalence of SI in patients prescribed these medications for weight loss relative to those treated for type 2 diabetes.
“This study is another source of evidence for patients and their providers to consider as they design their treatment plans and can serve to help end the potential bias and stigma of is for patients in need of GLP-1 medications.”
Interviewee
Jessica Probst
Principal, Real World Evidence, OM1
Jessica Probst, MPH is the Principal, Real World Evidence at OM1, where she is responsible for translating client needs for real-world data into scientifically robust, innovative evidence generation strategies. Jessica also serves as subject matter expert and thought leader for OM1’s Customer Success initiatives, providing epidemiologic support throughout the project lifecycle. Prior to joining OM1, she sat within the Real-World Solutions division of IQVIA, where she generated data-driven insights to inform occupational health and safety initiatives for major sports leagues. Jessica holds an MPH in Epidemiology from Emory University’s Rollins School of Public Health.
Disclaimer
The opinions expressed in this feature are those of the interviewee/author and do not necessarily reflect the views of The Evidence Base® or Becaris Publishing Ltd.