Pharmacovigilance in the GLP-1 Era: Safety Surveillance Challenges for Obesity Medicines and Real-World Use

1. How is the rapid uptake of GLP-1 receptor agonists for obesity reshaping traditional pharmacovigilance frameworks, particularly in terms of scale and speed of safety signal detection?

The rapid uptake of GLP-1 therapies is pushing pharmacovigilance beyond traditional, slower models built for smaller and more controlled treatment populations. Obesity use is broader, longer-term, and increasingly managed across primary care, specialist care, telehealth, and, at times, compounded-product channels. That means safety systems must detect signals earlier, interpret them faster, and track them across real-world settings. FDA has also highlighted underreporting concerns for some compounded GLP-1 products, reinforcing the need for more agile surveillance that combines spontaneous reports with active real-world data review and continuous signal evaluation.

2. What new methodological challenges arise when monitoring adverse events in diverse real-world populations using GLP-1 therapies, compared to controlled clinical trial settings?

Real-world monitoring is harder because GLP-1 use now extends far beyond tightly selected trial populations into patients with multiple comorbidities, polypharmacy, variable adherence, and different care pathways. Exposure can also be harder to characterise when treatment is obtained through obesity clinics, telehealth, or compounded channels, where dose escalation, product provenance, and storage quality may vary. Methodologically, that means pharmacovigilance teams must deal with greater confounding, missing data, indication misclassification, and uneven product quality while still detecting signals quickly and accurately.

3. Given the growing off-label and aesthetic use of GLP-1-based medicines, how can pharmacovigilance systems effectively differentiate and attribute risks between approved and unapproved indications?

The first step is to make exposure ascertainment much more disciplined. Safety systems should capture indication, product source, formulation, titration schedule, dose, and channel of access at case intake whenever possible. Approved branded use, pharmacy-compounded use, and illegally marketed or counterfeit products should not be pooled casually because the quality, dosing, and reporting environment differ materially. This matters even more because the FDA notes that adverse events from some compounded products are likely underreported and that fraudulent labeling and dosing errors have already complicated case interpretation.

4. How should regulators and manufacturers approach safety surveillance for compounded or non-regulated GLP-1 products that fall outside standard approval pathways?

They should take a risk-based, source-aware approach. Compounded and non-regulated products need intensified surveillance focused on product provenance, storage and shipping conditions, dosing practices, API quality, and counterfeit or fraudulent labeling. Regulators should continue rapid safety communications and targeted enforcement where quality concerns exist, while manufacturers should strengthen signal review for events that may reflect misuse, titration errors, or product substitution rather than the approved medicine itself. In parallel, encouraging MedWatch reporting is essential because underreporting can otherwise mask emerging risks in these channels.

5. What role do real-world data (RWD) and real-world evidence (RWE) play in identifying long-term and rare adverse effects of GLP-1 therapies, and what are their current limitations?

RWD and RWE are central because long-term and rare events often do not become fully visible in pre-approval trials. FDA explicitly recognises electronic health records, claims data, registries, and digital health sources as important inputs for post-market safety evaluation. Recent European safety reviews also show how real-world datasets can sharpen interpretation: PRAC used electronic health record studies in its review of suicidal ideation and used post-marketing surveillance, clinical trials, epidemiology, and literature in its semaglutide-NAION review. The main limitations remain confounding, incomplete exposure capture, misclassification, and uneven data quality, especially when compounded or unapproved products are involved.

6. How can signal detection algorithms be adapted or enhanced to manage the high volume of adverse event reporting associated with widely used obesity medications?

Signal detection needs to become more triaged and context-sensitive. High-volume GLP-1 reporting should be prioritised by verified product source, dose and titration pattern, seriousness, medical plausibility, and whether the event may reflect a known class effect, misuse, or a quality problem. Algorithms should also be designed to surface clusters linked to dosing errors, compounded products, or counterfeit supply, rather than treating all reports as equivalent. In practice, the strongest model is a hybrid one: automated prioritisation for speed, followed by medical review and continuous integration with real-world datasets for validation.

7. In the context of GLP-1 therapies, how can pharmacovigilance teams better assess causality when confounding factors such as comorbidities and polypharmacy are prevalent?

Causality is strongest when teams triangulate evidence rather than relying on single-source reports. That means combining case narratives with dose and timing information, dechallenge or rechallenge patterns, active-comparator analyses, and stratification by indication, baseline risk, and concomitant medicines. Real-world database studies can be especially valuable when they are carefully designed to handle confounding. Recent EMA reviews offer a good example: electronic health record studies helped contextualise suicidal ideation reports, while the NAION review drew on post-marketing surveillance, trials, epidemiology, and literature before reaching a regulatory conclusion.

8. What strategies can improve the monitoring and reporting of gastrointestinal, psychiatric, and metabolic adverse events commonly associated with GLP-1-based treatments?

Monitoring should become more structured at the point of care and at case intake. For gastrointestinal and metabolic events, pharmacovigilance systems should consistently capture dose-escalation stage, product source, indication, dehydration risk, concomitant medicines, and clinically relevant follow-up such as pancreatitis, gallbladder, renal, and hypoglycaemia evaluations reflected in product labeling. For psychiatric events, the priority is disciplined case documentation and follow-up rather than assumption of class causality, especially because FDA and EMA reviews did not find evidence supporting a causal association with suicidal thoughts based on available data.

9. How should pharmacovigilance frameworks evolve to address the increasing use of GLP-1 therapies in populations not traditionally studied, such as adolescents or individuals without diabetes?

Frameworks should become more population-stratified and less “one-size-fits-all.” FDA has approved Wegovy for adolescents aged 12 years and older with obesity, and the labeling shows that paediatric adverse-event patterns warrant dedicated attention, including gastrointestinal effects and gallbladder events. For broader non-diabetes use, pharmacovigilance should therefore analyse safety by age, indication, baseline metabolic status, growth and development considerations, concomitant treatments, and treatment duration, with longer follow-up than would normally be expected for short-course medicines.

10. What are the key challenges in communicating evolving safety data about GLP-1 drugs to both healthcare professionals and the public in a high-visibility therapeutic area?

The biggest challenge is communicating quickly without oversimplifying. In a high-visibility area, safety messages must clearly distinguish approved branded products from compounded or non-FDA-approved alternatives, and they must separate potential class questions from product-quality, dosing, or misuse issues. FDA’s recent communications illustrate how difficult that balance can be: one message addressed the removal of a suicidal-ideation warning after review of available data, while another warned about underreported and potentially unsafe non-approved compounded GLP-1 products. The communication burden is therefore not only scientific but also educational and trust-related.

11. How can digital health tools, wearables, and patient-reported outcomes be integrated into pharmacovigilance systems to enhance real-time safety monitoring for obesity medicines?

They should be used as fit-for-purpose adjuncts to traditional safety systems. The FDA describes digital health technologies as tools that can capture behaviour and physiology outside conventional clinic settings, and its RWE framework explicitly recognises digital health technologies as potential real-world data sources. For GLP-1 therapies, this supports near-real-time monitoring of adherence, dose escalation, gastrointestinal symptom burden, hydration-related issues, weight trends, and patient-reported tolerability. The key is to integrate these data with validated workflows, clear governance, and medical review so that signal speed improves without sacrificing interpretability or data quality.

12. What lessons can be learned from previous high-demand therapeutic classes (e.g., statins or COVID-19 vaccines) to strengthen pharmacovigilance practices in the GLP-1 era?

The main lesson is that volume changes pharmacovigilance itself. When a therapy class scales rapidly, organisations need stronger intake triage, clear case definitions, background-rate thinking, active surveillance alongside spontaneous reporting, and frequent public communication that can adapt as evidence evolves. FDA’s current GLP-1 communications and RWE activities point in that same direction: scalable surveillance, better use of real-world data, and transparent updates are essential when products are widely used, publicly discussed, and obtained through multiple channels. In that sense, the GLP-1 era rewards pharmacovigilance systems that are fast, layered, and highly communication-ready.

13. How should global regulatory agencies harmonise pharmacovigilance approaches for GLP-1 therapies, given differences in approval status, labeling, and market access across regions?

Harmonisation should focus first on methods, data standards, and signal governance, even when local approvals and labels differ. Agencies can align on common MedDRA terminology, interoperable ICSR transmission through ICH E2B(R3), and a shared aggregate benefit-risk structure through ICH E2C(R2)/PBRER, while still allowing region-specific appendices for local indications, access routes, or labeling decisions. In practice, the goal is not identical labels everywhere, but comparable evidence standards, faster signal sharing, and more consistent interpretation of class-wide risks. The recent FDA and EMA reviews on suicidality with GLP-1 medicines show that agencies can converge on core safety conclusions even through different procedural pathways.

14. Looking ahead, what innovations—such as AI-driven signal detection or decentralised data ecosystems—are most critical for ensuring robust safety surveillance in rapidly evolving therapeutic areas like obesity management?

The most important innovations are likely to be distributed data networks, fit-for-purpose real-world evidence, and carefully governed AI-assisted pharmacovigilance. FDA’s Sentinel Initiative already shows the value of a distributed data model for active post-market surveillance, and FDA’s broader safety-surveillance work now explicitly combines active and passive surveillance with RWE and AI-enabled approaches. For GLP-1 medicines, that kind of architecture is especially useful because exposure is large, real-world use is heterogeneous, and product channels may differ. The best future model is therefore not AI alone, but AI layered onto strong data standards, validated workflows, and medical review.

A second point is governance. Both FDA and EMA materials make clear that AI can support case management, prioritisation, and signal detection, but it does not remove the sponsor’s or regulator’s responsibility to validate models, monitor performance, and preserve explainability where it matters for decision-making. In rapidly evolving fields like obesity management, the winning combination will be decentralised data access, common data models, digital health inputs, and AI tools that accelerate triage without weakening scientific judgment.