Healthcare Utilization Among Chronic Rhinosinusitis Patients Using Glucagon‐Like Peptide‐1 Agonists

Chronic rhinosinusitis (CRS) is a multifactorial disorder of sustained sinonasal mucosal inflammation, often driven by dysregulated innate and adaptive immune responses [1]. Microbial pathogens and airborne irritants can trigger exacerbations of a complex cycle of mucosal inflammation, ciliary dysfunction, mucostasis, and biofilm formation [2]. Glucagon-like peptide-1 (GLP-1) agonists have recently gained popularity in the treatment of diabetes and obesity [3]. In addition to their effects on the gut axis, GLP-1 agonists have also been shown to exert systemic anti-inflammatory properties [4]. As such, the indications for GLP-1 agonists are being explored for the treatment of systemic inflammatory diseases, such as rheumatoid arthritis and inflammatory bowel disease, as well as autoimmune diseases [5]. Given the systemic immunomodulatory effects of GLP-1 agonists and the inflammatory disease processes underlying CRS, we aimed to explore the impact of GLP-1 agonists on healthcare utilization among patients with CRS. We hypothesized that patients with CRS who receive GLP-1 agonists have reduced healthcare utilization relative to those without GLP-1 agonists. These study findings aim to contribute to the evidence base on this novel drug class in sinonasal disease. This retrospective study assessed patients with a CRS diagnosis within the commercial Atropos health platform between 2015 and 2024. The Atropos platform is a national aggregated cohort of 46 million eligible deidentified patient records across the United States [6]. Patients were included if they had an International Classification of Diseases diagnosis code of CRS at an office visit, with a repeat diagnosis 3–12 months after index diagnosis. Patients were excluded if they were <18 years old or had follow-up <90 days. Patients who received a GLP-1 agonist in the office visit were included in the intervention arm, and patients who did not receive GLP-1 agonist anytime in their timeline were included in the control arm. Propensity-score analysis was performed matching age, sex, and Charlson comorbidity index, to compare patients who received GLP-1 agonists to those who did not. Outcomes of healthcare utilization associated with CRS diagnosis, including clinical encounters (emergency department [ED] visits, inpatient admissions, outpatient clinic visits), prescriptions, and rates of endoscopic sinus surgery (ESS) were measured 6–24 months after GLP-1 initiation, or 6–24 months after index CRS diagnosis in the non-GLP-1 cohort. A total of 220,479 patients with CRS were included: 11,020 (5.0%) received GLP-1 agonists, and 209,459 (95.0%) did not. The GLP-1 cohort had higher diabetes rates, hemoglobin A1c, and body mass index (BMI). On propensity-score analysis, 8055 patients were included in each cohort (Table 1). No differences in age, sex, allergy, asthma, or immunodeficiencies were observed on propensity-score analysis. The GLP-1 cohort had a higher incidence of private insurance and lower rates of Medicaid and Medicare Advantage plans (p < 0.001). The GLP-1 cohort had significantly lower rates of respiratory comorbidities such as bronchiectasis, cystic fibrosis, and chronic obstructive pulmonary disease, as well as lower rates of non-respiratory medical comorbidities (p < 0.001). The GLP-1 cohort had fewer antibiotic prescriptions (mean 2.9 ± 3.1 vs. 3.1 ± 3.2, p < 0.001) and fewer steroid prescriptions (2.1 ± 3.0 vs. 2.3 ± 3.2, p < 0.001). Biologic use was similar, with exception of dupilumab (0.88% in GLP-1 vs. 0.53%, p = 0.008). The GLP-1 cohort had lower rates of CRS-related ED visits (2.3% vs. 3.5%, p < 0.001), inpatient admissions (1.4% vs. 2.1%, p < 0.001), outpatient clinic visits (1.3 ± 2.1 vs. 1.5 ± 2.4, p = 0.006), and ESS (2.1% vs. 2.9%, p = 0.001; Figure 1). Further, a sub-analysis was performed among GLP-1 users comparing those on biologics (n = 390) to a propensity-matched cohort not on biologics (n = 390). Antibiotic prescriptions (3.7 ± 3.9 vs. 3.8 ± 3.8, p = 0.37), steroid prescriptions (3.5 ± 4.0 vs. 3.2 ± 4.5, p = 0.99), ED visits (3.9% vs. 3.1%, p = 0.56), and inpatient visits (1.5% vs. 2.6%, p = 0.31) were similar between GLP-1 users based on biologic use, but the biologic use cohort had higher rates of sinus surgery (4.6% vs. 1.5%, p = 0.013). To assess whether degree of weight loss impacted CRS outcomes, a sub-analysis of GLP-1 users was performed. Recipients of GLP-1 agonists with the highest tertile change in BMI (∆BMI) were compared to those with the lowest. The highest tertile ∆BMI had a mean weight loss of −28.7 lbs (standard deviation [SD] 20.1), whereas the lowest had a mean weight gain of +10.3 lbs (SD 15.7; p < 0.001). Despite significant differences in the amount of weight change among GLP-1 users, there was no difference in rates of sinus surgery (2.0% vs. 2.0%, p = 1.0), ED visits (2.5% vs. 2.1%, p = 0.40), or outpatient visits (1.5 ± 2.4 vs. 1.4 ± 2.2, p = 0.42), when comparing the highest versus the lowest tertiles of ∆BMI, respectively. Inpatient admissions were higher in the highest tertile ∆BMI of GLP-1 users compared to the lowest tertile (1.8% vs. 1.0%, p = 0.034). This study demonstrates that among a large national cohort, patients with CRS on GLP-1 agonists had lower rates of sinus surgery, antibiotic prescriptions, outpatient visits, ED visits, and inpatient admissions within a 6- to 24-month period of GLP-1 initiation. These findings were independent of weight loss and biologic use and suggest improved CRS disease control in patients on GLP-1 agonists. There may be several potential explanations for the reduced rate of surgical and medical interventions in the GLP-1 cohort. CRS with nasal polyps is mediated by a Th2-biased cytokine profile, which promotes eosinophilic inflammatory infiltrate. GLP-1 agonists have been shown to suppress the nuclear factor-kappa B pathway, which may reduce pro-inflammatory cytokines interleukin (IL)-5 and IL-13 that mediate the eosinophilic response [7]. The GLP-1 receptor pathway has also been shown to reduce tumor necrosis factor-α (TNF-α) signaling in immune cells [8], which may target the Th1-biased cytokine profile in CRS with nasal polyps. Whether the GLP-1 agonist pathway exerts effects through other mechanisms, such as ciliary function, reactive oxygen species, lipoprotein synthesis, and central nervous system, remains unknown [9]. GLP-1 agonists are also prescribed for non-diabetic patients with the goal of achieving weight loss, and obesity promotes a systemic pro-inflammatory state [10]. However, the lack of association between weight loss and GLP-1 use suggests that weight loss alone is not associated with reduced CRS-related healthcare utilization. Future studies are needed to examine the mechanistic etiologies underlying the global observation of reduced healthcare utilization in CRS with GLP-1 agonists. While the GLP-1 cohort had reduced burden of medical comorbidities, rates of specific comorbidities were approximately within 5% of the non-GLP-1 cohort. Rates of diabetes with and without complications were expectantly higher in the GLP-1 agonist cohort. Additionally, BMI was significantly higher in the GLP-1 cohort, both before and after propensity-score matching. Despite higher baseline comorbidities of diabetes and obesity, all healthcare utilization measures were significantly lower in the GLP-1 cohort. Of note, there were fewer antibiotic and steroid prescriptions in GLP-1 agonist recipients. Reduced antibiotic use may suggest improved reduction of acute infectious flareups with GLP-1 agonist usage. The GLP-1 cohort had a higher incidence of diabetes, and thus steroids may have been less likely to be prescribed due to steroid-induced hyperglycemia in diabetic patients. Future studies should assess the relationship between endoscopic sinus findings with GLP-1 agonist usage, such as signs of active sinus infection, mucosal edema, and polyp burden. This study has limitations inherent to national database analyses. While sinusitis diagnosis codes were linked to inpatient, outpatient, and ED visits, it is possible that the primary reason for the healthcare encounter was a reason other than sinusitis. Medication prescriptions were unable to be linked to diagnosis codes, and thus antibiotic and steroid prescriptions may reflect treatment for non-sinusitis etiologies. Insurance type distributions differed between cohorts, suggesting potential socioeconomic differences in access and disease management. Despite matching on Charlson Comorbidity Index, residual confounding from unmeasured social and clinical factors may persist. The underlying mechanism of reduced healthcare utilization among GLP-1 agonist recipients with CRS should be further explored, particularly with regard to longitudinal observations, endoscopic findings, and potential confounding factors. The authors declare no conflicts of interest. The authors have nothing to report.