As GLP-1 receptor agonists become among the most prescribed drug classes in U.S. history, a fundamental disconnect is emerging: the populations taking these medications in clinical practice look very different from those enrolled in the landmark trials that established their efficacy. That gap — and what to do about it — was the central concern of a May 2025 expert convening organized by the National Institute of Diabetes and Digestive and Kidney Diseases.
The workshop brought together regulators, payers, guideline authors, and academic researchers to interrogate how real-world evidence (RWE) — drawn from electronic health records, insurance claims, pharmacy databases, and patient registries — can fill what randomized trials structurally cannot answer. Key knowledge gaps identified include long-term cardiovascular and renal safety beyond trial follow-up windows, comparative effectiveness across racial and socioeconomic subgroups, outcomes in pediatric populations, and the clinical consequences of high discontinuation rates that routinely exceed 50% in observational datasets. Compounding the challenge, the heterogeneous and rapidly shifting coverage landscape across Medicare, Medicaid, and commercial insurers creates what researchers termed "time-varying confounding" — a methodological problem that makes it difficult to isolate drug effects from access effects in administrative data.
This workshop synopsis is best understood as a state-of-the-field audit rather than a new empirical finding. Its significance lies in codifying where GLP-1 science is structurally weakest just as clinical adoption is accelerating fastest. The persistent underrepresentation of pediatric and underserved populations in both trials and high-quality real-world datasets is a particularly under-discussed vulnerability — one that matters enormously given that obesity disproportionately affects lower-income and minority communities, precisely the groups most likely to face coverage barriers and medication discontinuity. While RWE methodologies are maturing, the workshop's frank acknowledgment of unresolved confounding, incomplete medication capture due to compounded formulations and off-label use, and data reliability limitations signals that regulatory and clinical confidence in RWE for GLP-1 decisions should still be calibrated with considerable caution. This is an important institutional checkpoint, not a resolution.