GLP-1 Safety Signals in 2024: What the FAERS Data Actually Tells Us

When millions of people start taking a drug class within a few years, the real-world safety picture inevitably diverges — sometimes subtly, sometimes significantly — from what randomized trials captured. That divergence is exactly what pharmacovigilance databases exist to detect. The FDA Adverse Event Reporting System (FAERS) now contains nearly 18 million case reports, and researchers have been mining it aggressively to understand what GLP-1 receptor agonists are actually doing to people outside the controlled conditions of a clinical trial. The findings from 2024 and early 2025 are nuanced, occasionally surprising, and worth understanding if you or someone you care about is using semaglutide, liraglutide, dulaglutide, or any other agent in this class.

This is not a story about whether GLP-1 medications work. The evidence on weight loss and metabolic improvement is robust. This is a story about signal detection — the epidemiological science of identifying patterns in post-marketing data that warrant closer attention. Some signals are strong. Some are weak. Some findings are reassuring. Understanding the difference is what separates informed decision-making from either panic or blind confidence.

The Gastrointestinal Picture: Well-Known, Still Worth Watching

Anyone who has spent time on semaglutide knows that nausea, vomiting, and constipation are not rare occurrences. The FAERS data confirms what clinical trials already suggested, but real-world analysis adds texture that trial populations often miss. systematically characterized gastrointestinal safety signals across the drug class, reinforcing that GI events remain the dominant adverse event category by volume. What makes real-world pharmacovigilance valuable here is that it captures the full severity spectrum — from mild nausea managed with dietary adjustments to hospitalizations for severe vomiting or gastroparesis — and it captures populations that trials often exclude, including older adults, people with complex comorbidities, and those on polypharmacy regimens.

For men using GLP-1 medications alongside an active training program, the GI burden has practical consequences that go beyond discomfort. Inadequate caloric intake driven by nausea and early satiety, combined with the anorexigenic effect of the medication itself, creates a real risk of insufficient protein consumption and accelerated lean mass loss. This is not a theoretical concern. If you are taking a GLP-1 agent and strength training, tracking protein intake — aiming for at least 0.7 to 1 gram per pound of bodyweight — is not optional. It is the primary nutritional strategy for preserving the muscle you have built while the medication does its metabolic work.

Separately, examined metabolic and nutritional adverse events associated with GLP-1 receptor agonists across the FAERS database, highlighting that the downstream nutritional consequences of these drugs deserve dedicated monitoring — not just the acute GI symptoms, but the longer-term metabolic implications of chronic appetite suppression and dietary restriction. Clinicians and patients alike should treat nutritional adequacy as an active management goal, not a passive afterthought.

The Eye, the Brain, and the Signals That Demand Deeper Investigation

The more unexpected findings from recent pharmacovigilance work involve systems that most patients — and arguably many prescribers — are not routinely monitoring. Two areas stand out: ocular health and neuropsychiatric effects.

On the ocular front, the data are striking. A 2025 study published in Expert Opinion on Drug Safety analyzed FAERS reports from 2004 to 2024 and found that semaglutide and lixisenatide were significantly associated with ocular adverse events, with reporting odds ratios of 1.25 (95% CI 1.20–1.31) and 1.96 (95% CI 1.70–2.27) respectively. The most commonly reported events included blurred vision, visual impairment, and diabetic retinopathy. Perhaps most clinically important: some ocular adverse events appeared as early as 10 days after treatment initiation. The researchers used disproportionality analysis — a validated pharmacovigilance methodology — alongside gene enrichment analysis to explore potential biological mechanisms linking GLP-1 receptor activity to ocular tissue.

This finding matters for a specific reason. Rapid improvements in blood glucose control, which GLP-1 medications can produce, are already known to be associated with early worsening of diabetic retinopathy — a phenomenon called early worsening retinopathy following glycemic improvement. But the FAERS signals here extend beyond diabetic populations and suggest the relationship between GLP-1 receptor agonism and ocular health may involve additional pathways. For anyone starting a GLP-1 medication, particularly those with pre-existing diabetes or any history of retinal issues, establishing a baseline with an eye care provider before or shortly after initiation is a reasonable precaution that this data supports.

The neuropsychiatric picture is more complex and requires careful interpretation. Two studies point in somewhat different directions, and both deserve attention rather than selective citation. A 2025 pharmacovigilance study in EClinicalMedicine analyzed FAERS and WHO VigiBase data through December 2024 and found that semaglutide — but not liraglutide or tirzepatide — demonstrated statistically significant signals of disproportionate reporting for depressive disorders in both databases (FAERS ROR 1.26, 95% CI 1.15–1.37; VigiBase ROR 1.38, 95% CI 1.27–1.49). The temporal analysis suggested an early failure pattern, meaning the association appeared relatively early in treatment rather than accumulating over time. Importantly, stratified analyses found increased disproportionality in females and in reports submitted by healthcare professionals, lending some credibility to the signal quality.

This finding is counterintuitive given the body of evidence suggesting GLP-1 receptor agonists may have neuroprotective and even antidepressant properties. The authors themselves acknowledge this inconsistency and frame it carefully: the signal is semaglutide-specific, not a class-wide phenomenon, and pharmacovigilance data cannot establish causation. Confounding is a real concern — people initiating weight-loss medications are often doing so during periods of metabolic and psychological stress, and depression is common in this population regardless of medication. Nevertheless, a signal that appears consistently across two independent international databases warrants monitoring rather than dismissal.

On the question of suicidal and self-injurious behaviors — which drew significant media attention and regulatory scrutiny in 2023 — the pharmacovigilance data are considerably more reassuring. A 2024 study published in BMC Medicine analyzed FAERS data from 2018 through 2022 and identified 204 cases of suicidal or self-injurious behaviors associated with GLP-1 receptor agonists. After disproportionality analysis, the investigators found no signal — meaning the reporting rate for these events was not higher than what would be expected given the baseline rate in the underlying population. Of the 204 cases, 81 involved co-medication with antidepressants, antipsychotics, or benzodiazepines, suggesting that pre-existing psychiatric conditions were likely driving many of these reports. This is methodologically important: the absence of a disproportionate signal in a large database is meaningful evidence, even if it is not absolute exoneration.

The practical implication is not that neuropsychiatric monitoring is unnecessary, but that it should be targeted and proportionate. For men starting a GLP-1 medication who have no personal or family history of depression or mood disorders, the absolute risk appears low. For men with pre-existing psychiatric conditions or who are already on psychotropic medications, closer monitoring is warranted — not because the drugs are proven harmful in this population, but because the pharmacovigilance literature identifies it as a higher-risk subgroup.

How to Use This Information Practically

Pharmacovigilance data is hypothesis-generating, not hypothesis-confirming. That distinction is not a caveat designed to minimize the findings — it is a fundamental feature of how signal detection works. FAERS captures spontaneous reports, which means it overrepresents severe events, is subject to reporting bias, and cannot control for confounders the way a randomized trial can. What it does exceptionally well is identify patterns at population scale that would never emerge in the controlled environment of a 6-month clinical trial enrolling highly selected participants.

The actionable framework from this body of research is relatively straightforward. If you are using a GLP-1 medication, establish a baseline for the organ systems now flagged by pharmacovigilance data: get your eyes checked, particularly if you have diabetes or any prior retinal history. Tell your prescriber about any changes in mood, energy, or cognition — especially in the first weeks of treatment, given the early failure pattern observed in the depression signal data. Do not assume that GI symptoms are automatically manageable; persistent vomiting severe enough to disrupt fluid and electrolyte balance deserves clinical evaluation, not just dietary adjustment. And maintain nutritional vigilance, because the appetite suppression that makes these drugs effective is also the mechanism through which nutritional deficiencies and lean mass loss can occur if you are not actively countering them.

For men not on GLP-1 medications, this research still carries a lesson. The FAERS database and pharmacovigilance methodology represent the best available tool for detecting real-world drug safety signals at scale, and the 2024 data collectively paints a picture of a drug class that is broadly safe but not without meaningful risks in specific organ systems and specific populations. As prescribing continues to expand across both diabetic and non-diabetic populations — and as new formulations and indications reach the market — the signal landscape will continue to evolve. Staying informed is not paranoia; it is applied health literacy.

The Takeaway

The 2024 pharmacovigilance literature on GLP-1 receptor agonists delivers a message that resists simplification: these are powerful, broadly effective medications with a safety profile that is mostly well-characterized but contains several signals demanding continued scrutiny. Ocular adverse events, particularly with semaglutide, appeared earlier and more frequently than many clinicians anticipated. A semaglutide-specific signal for depressive disorders emerged consistently across two major international surveillance databases, even as the same class shows potential neurological benefits in other contexts. Suicidal behavior signals, by contrast, did not reach statistical significance — a finding that should offer genuine reassurance to patients and prescribers who were concerned by earlier media reports. And the gastrointestinal and nutritional consequences of long-term GLP-1 use remain the highest-volume safety concern by far, with meaningful implications for anyone trying to simultaneously manage body composition and metabolic health. Whether you are on one of these medications or simply trying to understand the landscape, the research rewards careful reading over headline-driven conclusions.