Investigators queried the FAERS database from 2018Q1 through 2024Q4. They included individual case safety reports where semaglutide, dulaglutide or empagliflozin was listed as the primary suspect. Panic attack cases were identified using the MedDRA preferred term. The dataset contained 33,267 semaglutide reports, 54,282 dulaglutide reports and 25,380 empagliflozin reports; recorded panic attacks numbered 87, 28 and 13 respectively. Those crude counts produced large unadjusted reporting odds ratios (RORs) for semaglutide versus the two active comparators—around fivefold higher. Short sentence: crude disproportionality signaled a possible association.
After multivariable adjustment for age, sex and indication (type 2 diabetes, obesity/weight management), the signal attenuated. The adjusted reporting odds ratios (aRORs) fell to approximately 2.8 versus dulaglutide and 2.75 versus empagliflozin. The authors reported adjusted 95% confidence intervals; these aRORs were smaller than the crude estimates, indicating confounding by measured covariates influenced the initial observations. The report also included a sensitivity analysis that used all other primary-suspect drugs in FAERS as the reference set; against that broad reference the overall aROR was 1.48.
These analyses were framed as disproportionality screening. The authors stressed that such analyses are hypothesis-generating, cannot provide incidence, and cannot establish causality. The dataset and methods reported in the source do not provide individual-level clinical adjudication beyond what is present in spontaneous reports.
The study team performed prespecified temporal assessments to evaluate notoriety bias. They compared reporting before July 1, 2023 with reporting after that date, and they contrasted 2018–2022 with 2023–2024. Reports of panic attack per 10,000 semaglutide reports rose from 6.7 in 2018–2022 to 42.0 in 2023–2024. A drug-by-period interaction reached statistical significance (p = 0.007). These temporal patterns coincided with heightened media and regulatory attention to mental health outcomes for GLP-1 products during 2023.
When the authors restricted analyses to reports received before July 1, 2023, crude RORs were lower (about 2.2 versus dulaglutide and 2.15 versus empagliflozin) and attenuated further after adjustment to non-significant levels (aRORs around 1.14–1.38, with confidence intervals crossing unity). In that pre-warning window, using all other primary-suspect drugs in FAERS as the comparator yielded an aROR below one (0.76). Short, concrete point: the surge in reporting occurred after mid-2023 and is temporally aligned with external attention, a pattern consistent with increased reporting rather than clear emergence of a drug-induced event.
The authors used reporting proportions and time-stratified analyses to interrogate the signal. They documented that the pattern of rapid increase after mid-2023 is compatible with notoriety bias—that is, heightened awareness leading to greater voluntary reporting—rather than definitive evidence of a new pharmacologic adverse effect.
Parallel to the pharmacovigilance analysis, the group conducted a systematic review of the literature to November 24, 2025. They screened randomized trials, cohort and cross-sectional studies, pharmacovigilance analyses and case reports that reported panic attacks or acute anxiety outcomes with semaglutide. Twelve studies met inclusion criteria according to the article.
Across the identified randomized trials and active-comparator observational cohorts, the investigators did not find a consistent signal linking semaglutide to increased panic attacks or anxiety-related outcomes. Sentence with contrast: clinical trial pools and well-controlled cohort comparisons did not reproduce the disproportionality finding in a consistent way. The source document did not report pooled effect estimates from these trials in the abstract, nor did it provide additional trial-level numerical details in the abstract text; readers must consult the full text for study-by-study metrics.
Case reports and spontaneous reports were part of the evidence base, but the review emphasized heterogeneity in outcome definitions and limited ability of trials to detect rare or idiosyncratic events. Short sentence: randomized data did not consistently support a pharmacological association.
The authors concluded that crude FAERS analyses suggested a semaglutide–panic attack signal but that the association attenuated after covariate adjustment and was absent before the July 2023 regulatory communication. The marked rise in reports after mid-2023, together with the temporal interaction, supports a role for notoriety bias in inflating spontaneous-report counts. Triangulation with randomized and active-comparator observational evidence did not show a consistent pharmacological effect across study types.
The authors spelled out important limitations that affect interpretation. Disproportionality measures cannot estimate incidence or prove causality. The FAERS dataset contains voluntarily submitted reports that lack standardized adjudication and are subject to reporting biases. The source abstract did not provide granular trial-level effect sizes or patient-level data; those details require consulting the full text (PMCID: PMC13713200). The study did adjust for age, sex and indication but residual confounding and misclassification within spontaneous reports remain possible.
Unanswered clinical questions documented in the source include whether rare, individual-level susceptibility to acute anxiety exists with semaglutide, and whether specific patient subgroups might be at higher risk. The timing of the spike in reports after mid-2023 raises a regulatory milestone: the July 2023 communication demarcated the pre- and post-awareness periods used in the analysis. Future post-marketing work that the authors implied would be useful includes prospectively designed observational cohorts or adjudicated safety registries focused on neuropsychiatric outcomes to clarify incidence and causality. The report declared no conflicts of interest.
End point: the source emphasizes cautious interpretation—disproportionality analysis can flag signals, but it cannot on its own establish that semaglutide causes panic attacks. Future targeted studies and adjudicated safety monitoring will be required to resolve the remaining uncertainties.