Fewer than a third of commercially insured US adults who started a GLP-1 drug for obesity were still taking it consistently a year later, an analysis of 201,229 claims records has found. About 64% fell into one of three discontinuation patterns: stopping almost immediately, stopping within a few months, or tailing off later in the year. The study, published on 17 September in Diabetes, Obesity and Metabolism, cannot say why anyone stopped. What it does is put a number on something clinics have described anecdotally since these drugs arrived.
What the claims data show
Researchers at the University of Florida, Purdue and Indiana University used MarketScan commercial claims from 2021 to 2023, selecting everyone aged 10 or over who had obesity, did not have type 2 diabetes, and started a GLP-1 receptor agonist. Adherence was measured as the proportion of days covered — the share of each month for which the person had drug dispensed to them — tracked across 12 months. Group-based trajectory modelling then sorted people by the shape of that curve rather than by a single pass/fail cutoff.
Among 201,229 adults, five patterns emerged: immediate discontinuers (19.6%), early discontinuers (23.5%), late discontinuers (20.7%), intermittent adherents (5.5%) and persistent users (30.7%). Mean annual proportion of days covered ranged from 0.10 in the first group to 0.87 in the last. The largest single group, just under a third, stayed on treatment; roughly 64% sat in one of the three discontinuation trajectories.
The 1,139 paediatric patients split more evenly, each of the five groups holding roughly 18% to 22%, with persistent users at 20.3%. Region mattered in both cohorts: compared with the South, people in the Northeast were more likely to be persistent users, with an odds ratio of 1.32 in adults and 1.65 in paediatric patients.
What it does not explain
A claims database records what was dispensed, not what happened. It cannot separate someone who stopped because the side effects were intolerable from someone whose employer changed plans, whose prior authorisation lapsed, who ran into a supply shortage, or who reached a weight they were content with and chose to stop. The 2021–2023 window covers exactly the period when supply of semaglutide and tirzepatide was erratic and coverage rules were tightening, so part of this curve is likely plumbing rather than patient choice.
The regional gap is the most suggestive finding and the least explained. An odds ratio of 1.32 between Northeast and South, holding in both adults and children, is the sort of pattern that usually tracks insurance design and local coverage policy rather than anything about the patients themselves. The study did not test that. What the result does establish is scale: over a year, discontinuation is not an edge case in this population but the majority experience — and that is among people with commercial insurance, the group best placed to afford treatment. For anyone weighing what these drugs cost over time, the operative question is less the monthly price than how long people manage to keep paying it.
Everyone in this cohort had commercial insurance, so the findings say nothing about Medicare, Medicaid or cash payers, who face different economics. The window predates the 2026 pricing changes and the newer self-pay channels, either of which could move these numbers. The design is observational and retrospective, with no data on side effects, weight outcomes, or any individual’s reason for stopping — the trajectories describe behaviour, not motives. And the paediatric cohort, at 1,139 patients, is small enough that its five near-equal groups should be read as provisional.
Sources: Huang W, Kotecha P, Kimmel SE, et al. “Real-World Adherence Trajectories of GLP-1 Receptor Agonists Among Adults and Paediatric Patients With Obesity.” Diabetes, Obesity and Metabolism, published online 17 September 2026 (doi:10.1111/dom.71299).
