Use of GLP-1 Receptor Drugs Linked to Excessive Sleepiness in Real-World Patients

Glucagon-like peptide-1 receptor agonists (GLP-1RA) have become a cornerstone therapy in the treatment of type 2 diabetes mellitus (T2DM), offering glycemic control, weight reduction, and favorable cardiovascular profiles. More recently, agents such as semaglutide and liraglutide have received FDA approval for obesity treatment in individuals without diabetes, expanding on previous indications [1].

While the peripheral metabolic actions of GLP-1RAs are well established, accumulating evidence suggests that GLP-1RAs exert significant central nervous system (CNS) effects. GLP-1 receptors are widely expressed throughout the brain, including regions such as the hypothalamus and brainstem nuclei [2–4]. GLP-1 signaling has been shown to suppress ghrelin, a hunger promoting hormone that stimulates orexin producing neurons in the lateral hypothalamus [5]. Orexins have a critical role in maintaining arousal and stabilizing sleep wake transitions [6]. Preclinical studies suggest that GLP-1RAs may interact with this system by dampening orexinergic activity, potentially leading to hypersomnolence, altered REM dynamics, and other sleep-related disturbances [6, 7].

Despite these mechanistic insights, the effects of GLP-1RAs on human sleep physiology remain poorly understood. Little is known about how these agents are associated with hypersomnolence, parasomnia, or narcolepsy-like symptoms in the broader population [8]. Post-marketing pharmacovigilance analyses have identified neurological adverse event signals associated with GLP-1RA use, including reports of lethargy and central mediated symptoms [9]. This study aims to investigate the association between GLP-1 RA use and the onset of hypersomnolence, parasomnia, and related outcomes at 1- and 5-year timepoints, using a large real-world dataset of patients.

Materials & Methods.

After application of the inclusion and exclusion criteria, 173,446 individuals were identified for the experimental group (GLP-1 RA users) and 2,611,755 for the control group. After 1:1 propensity matching, there were two equal cohorts of 118,993 patients for use in the outcome analysis. Baseline characteristics were well balanced with standardized mean differences < 0.10 across all covariates. The mean age was approximately 39 years old for both cohorts and ~ 62% female population. Demographic data and baseline characteristics for the population studied are listed in Table 1.

This study demonstrated a significant association between GLP-1RA use and the development of hypersomnolence over both short- and long-term follow-up periods. These findings are consistent with an emerging body of literature suggesting that GLP-1RAs, influence central nervous system pathways involved in sleep regulation. GLP-1 receptors are expressed in the lateral hypothalamus and brainstem nuclei, areas that regulate arousal, REM/non-REM transitions, and circadian rhythms [2–4, 10, 11]. The lateral hypothalamus contains orexin-producing neurons that maintain wakefulness; disruption of this system is a known mechanism in narcolepsy. It is therefore plausible that exogenous GLP-1RA exposure may interfere with orexinergic tone, destabilizing sleep-wake homeostasis, and leading to hypersomnolence and, as observed in our study [11].

Supporting this hypothesis, Khan and Doty reported that GLP-1RAs impair taste function via central pathways, providing indirect evidence of their influence on neurological signaling [12]. Additionally, a system-based review by Fadel et al. emphasized that GLP-1RAs may exert broader neurocognitive and sleep-related effects that remain underexplored [13]. Patient-reported data also reinforce this profile as shown in a systematic review of GLP-1RA experiences. Ibsen et al. documented frequent complaints of fatigue, drowsiness, and poor sleep quality symptoms often underrecognized in clinical settings [14].

While earlier studies have primarily focused on gastrointestinal or metabolic safety profiles such as pancreatitis risk [15] or delayed gastric emptying with aspiration [16] the potential neurologic and sleep-related adverse effects remain poorly characterized. Aldhaleei et al., for instance, found increased gastrointestinal (GI) adverse events in GLP-1RA users, which could contribute to nocturnal reflux and sleep disruption [17]. These findings collectively suggest that GLP-1RAs may influence multiple pathways relevant to sleep, including central neurochemical balance and peripheral GI function.

Although the HR for hypersomnolence indicates a 21–23% relative increase in hazard among the GLP-1RA cohort, the absolute risk was modest. At 1-year the cumulative incidence difference was ~ 0.4%, and at 5-years was ~ 0.7%. This distinction is clinically relevant, as the RR measures may overstate perceived impact when the baseline events are low. Accordingly, the observed association should be interpreted in the context of a small absolute increase in risk, particularly when weighted against the established metabolic and cardiovascular benefits of GLP-1RA therapy.

Overall, our findings add to a growing recognition of the neurophysiological consequences of GLP-1RA therapy. The potential for these agents to disrupt sleep-wake stability through modulation of orexin signaling and REM architecture warrants further mechanistic investigation in future studies.

In our analysis, we also found that GLP-1RA use was significantly associated with increased incidence of iron deficiency at the 5-year endpoint. Prior studies corroborate this relationship, potentially through altered GI absorption or inflammatory modulation [18–20]. Specifically, the mechanistic links between iron deficiency and both reduced absorption and chronic inflammation are often related through the regulatory hormone hepcidin [21]. While obesity can reduce iron absorption through increased hepcidin concentrations [22]. Given the known link between iron deficiency and RLS, this may represent a key pathophysiologic pathway contributing towards the elevated RLS risk observed in our study. RLS is known to contribute to disturbed sleep and affects nearly all patients who are seeking treatment [23, 24]. Therefore, GLP-1RA induced iron deficiency may represent a mechanistic pathway through which these agents contribute to RLS and a downstream sleep disruption.

Our study carries important clinical significance given the limited understanding of sleep related consequences from GLP-1RA use. As these agents gain traction not only in individuals with metabolic disease but also in otherwise healthy populations, their clinical reach continues to expand. Concurrently, as they are increasingly explored for neurologic indications, recognition of potential adverse effects such as somnolence, parasomnia, and sleep fragmentation becomes essential [25, 26]. These findings may aid clinicians to anticipate and manage these side effects earlier, especially in patients with baseline hypersomnia or sleep related disorders. Additionally, this knowledge can help to guide more personalized and informed risk benefit assessments when considering GLP-1RA therapy in clinical practice.

Though our results are statistically and clinically significant, there are several limitations to this study that must be acknowledged. First, the study relied on de-identified administrative data which precluded access to patient level data such as symptom severity, medication adherence, or objective sleep architecture metrics from polysomnography.

Secondly, given the retrospective cohort design, causal inference cannot be established despite rigorous propensity score matching and time to event analysis. Unmeasured confounding variables such as socioeconomic status, environmental sleep conditions, or unrecorded psychiatric comorbidities may have influenced the observed associations. There is also the consideration of diagnostic coding errors or omission that could have introduced misclassification bias. As with all retrospective observational studies, the possibility of confounding by indication or other systemic differences between groups cannot be fully excluded. This analysis also did not differentiate between specific GLP1-RA agents, dosages, duration of treatment, or adherence patterns. The observed association between GLP-1RA use and iron deficiency also raises the possibility of mediation effects. Iron deficiency is a known contributor to RLS and sleep fragmentation. Hence, some sleep related findings may be due to indirect pathways rather than CNS effects of GLP-1RA therapy. Our study also used polysomnography as an inclusion criterion which likely selected a population already undergoing evaluation for sleep disturbances limiting generalizability.

However, this study has several strengths. To our knowledge, this is the first large scale study evaluating GLP-1RA in relation to hypersomnolence and related outcomes using real world data. We utilized a large, multicenter population with diverse demographics which enhances generalizability. Our study also reduced baseline bias via propensity score matching, and time to event analysis provided insight to early and delayed risks.

In conclusion, our study found that GLP-1RA use was associated with a statistically significant increase in hazard of hypersomnolence at both 1- and 5-year endpoints. Although the relative increase in hazard was consistent, the absolute risk increase was modest, remaining below 1% over 5 years. An increased hazard of iron deficiency at 5-years and decreased hazard of Parkinsonism at 5-years were observed, though the latter finding was based on low even counts and should be interpreted with caution. Future research should include prospective studies with polysomnographic data, pharmacodynamic exploration of CNS effects, and stratification by psychiatric and neurologic comorbidities. As the role of GLP-1RA therapy continues to expand, it is critical to understand the broader impact on sleep, cognition, and underlying mechanisms of these associations.

Below is the link to the electronic supplementary material.

CRediT: Louie Kamel-Abusalha: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Project administration, Writing – original draft, Writing – review & editing; Ahmed M Afifi: Conceptualization, Data curation, Formal Analysis, Investigation, Methodology, Resources, Software, Visualization, Writing – original draft, Writing – review & editing; Abdallatif Dawoud: Writing – original draft; Zumair Hayath: Writing – original draft; Muhammad F Bouso: Writing – original draft; Andre Aguillon: Writing – review & editing; Ragheb Assaly: Conceptualization, Project administration, Supervision, Validation, Writing – review & editing.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

The data that support the findings of this study were obtained from the TriNetX Research Network, a federated, multi-institutional clinical research platform that provides access to de-identified electronic health record data. Due to data use agreements and patient privacy protections, these data are not publicly available and cannot be shared directly by the authors. Researchers with appropriate institutional approvals may access the data through the TriNetX platform (https://www.trinetx.com↗) subject to TriNetX policies and regulatory requirements.

Ahmed M. Afifi, Email: Ahmed.Abdelwahab@UToledo.Edu

Ragheb Assaly, Email: Ragheb.Assaly@utoledo.edu.