Considerations for evaluating the impact of GLP‐1 receptor agonist use on inpatient outcomes

Glucagon-like peptide-1 receptor agonist (GLP-1RA) and co-agonist use for cardiometabolic risk reduction represents one of the most significant therapeutic shifts in modern medicine. As Food and Drug Administration (FDA) approval expands and costs decline, GLP-1RA use will expand, and inpatient clinicians will increasingly encounter patients exposed to them. How will the widespread use of GLP-1RAs affect inpatient care? In their article, “Prior GLP-1 Agonist Use Is Not Associated with Adverse Inpatient Critical Care Outcomes: A Propensity-Matched Analysis,”1 Park et al. provide an early attempt to address this evidence gap. The authors examine associations between prehospitalization GLP-1RA exposure and intensive care unit (ICU)-related outcomes, finding no differences in intensive care unit (ICU) mortality, ICU length of stay (LOS), or hospital LOS for patients receiving a GLP-1RA prescription in the 12 months before admission.1 While reassuring, there are several limitations to the study design that should temper interpretation, including (1) challenges in identifying GLP-1RA exposure, (2) potential for collider bias, and (3) inability to measure the hypothesized pathway of muscle loss. First, accurately capturing GLP-1RA use in retrospective data is difficult. Key details (e.g, dose, titration, and duration) are typically unavailable, despite influencing the magnitude of GLP-RA's physiological effects. There is poor persistence among GLP-1RA users, with 50%–80% discontinuing therapy within 1 year.2 Further, many patients receive but never fill prescriptions due to a lack of insurance coverage or high costs. The authors did include a sensitivity analysis, limiting prescriptions to within 2 months, bringing exposure and outcome into closer temporal proximity. In this subanalysis, patients on GLP-1RAs had slightly longer ICU LOS.2 However, this alone does not address the risk of exposure misclassification.3 Second, the chosen analytic approach is vulnerable to collider bias, or a misleading association when a study includes only people who share a common downstream event. In this case, ICU admission is a “collider” because it is influenced by both GLP-1RA use and other risk factors for disease severity. This selection process can make exposure appear harmful or beneficial even when no causal relationship exists. For example, if GLP-1RAs theoretically decrease ICU admissions due to positive cardiometabolic effects, then GLP-1RA users who reach the ICU may be sicker, frailer, or less responsive to therapy. This would make GLP-1RAs appear harmful or neutral when they could, in fact, be protective when looking at the larger population of hospitalized patients. Third, this study is based on the premise that GLP-1RAs increase myopathy, which is commonly experienced during hospitalization due to acute illness and prolonged immobility. GLP-1RA-associated weight loss does reduce fat-free mass, with 25%–50% of total weight loss coming from lean tissue, including skeletal muscle.3, 4 However, the clinical significance of these effects remains uncertain. Initial rapid weight loss may exacerbate sarcopenia, particularly for those at highest risk (e.g., elderly and postmenopausal patients), but data suggest GLP-1RAs contribute to adaptive improvements in muscle quality over time, leading to protective effects.5 Park et al. could not assess prehospitalization weight trends, fat-free mass, or functional strength—understandable limitations given such measures are not routine during hospitalization. As a result, this hypothesized pathway linking GLP-1RAs to ICU outcomes could not be evaluated nor confirmed. Until stronger evidence emerges, any theoretical risk of myopathy must be qualified based on actual weight loss experiences and considered alongside the substantial cardiometabolic benefits of GLP-1RAs. With these limitations, the findings of Park et al. should not be interpreted as establishing the causal relationship between GLP-1RA exposure and ICU outcomes. However, these results highlight the need for more study of how GLP-1RAs impact hospitalization. Future research should examine outcomes sensitive to changes in metabolic status, muscle mass, and functional reserve (e.g., glycemic control, wound healing, infection risk, grip strength, and rehab needs) to identify positive and negative effects of these drugs. Additionally, as only a minority of hospitalized patients require ICU care, the most immediate—and practical—questions for hospitalists will arise at the floor level. For example, when should GLP-1RAs be resumed during or after admission? How should clinicians retitrate therapy after missed doses, given the months-long escalation needed to minimize side effects? What instructions and prior authorization are required at discharge to prevent treatment interruptions? And for patients with diabetes, how should glycemic control be maintained when GLP-1RAs are temporarily held? These routine but nuanced issues underscore the need for knowledge sharing across clinical specialties. Obesity is the most prevalent chronic illness in the United States. As GLP-1RA and co-agonist use accelerates—and new weight-modifying agents emerge—the inpatient setting will increasingly intersect with obesity pharmacotherapy. This shift presents an opportunity to build inpatient obesity medicine expertise and develop clinical pathways to ensure consistent care across hospital systems. Hospitalization must also be leveraged to identify patients eligible for therapies and coordinate follow-up with obesity medicine specialists. As such, hospitalists will play an important role in understanding and managing the inpatient implications of paradigm-shifting obesity medications. The authors have no funding to report. Jessica L. Schwartz received grant funding from Novo Nordisk Inc for obesity-related research. This funding relationship has ended within the past 2 years. Michelle Ogunwole is supported by the National Institute of Diabetes and Digestive and Kidney Diseases through the DiabDocs K12 program (K12DK133995), the Robert A. Winn Clinical Investigator Leadership Award in Cardiovascular Research (Winn CILA-CV) from the American Heart Association, and the Johns Hopkins Center for Innovative Medicine. These sources had no role in this work, and the author reports no relevant conflict of interest.