Weight trajectories after last Tirzepatide or Semaglutide prescription across a federated health network

Technical Summary: Weight Trajectories After Last Tirzepatide or Semaglutide Prescription

Problem Statement
While GLP-1 receptor agonists (GLP-1RAs) like semaglutide and tirzepatide have demonstrated significant efficacy in weight loss, prior literature indicates that discontinuation often leads to weight regain and a reversal of cardiometabolic benefits. However, the real-world association between the cessation of GLP-1RA prescriptions and subsequent weight change remains under-explored. A critical gap exists in understanding patient trajectories following the last prescription by a primary care or specialist physician, particularly given the increasing complexity of patient access to these medications through non-traditional sources (e.g., online providers, medical spas). This study addresses the question of whether weight loss is universally reversed after the last documented prescription or if subsets of patients maintain durable weight loss.

Methodology
This was a retrospective observational study utilizing de-identified Electronic Health Record (EHR) data from a federated network of tertiary clinical centers in the United States, accessed via the nference nSights Analytics Platform.

• Cohort Selection: The study included 4,182 patients who received at least three prescriptions for semaglutide or tirzepatide within a one-year period (08/01/2019 – 07/01/2025), with prescriptions spaced no more than four months apart. Patients were required to have at least six months without a subsequent GLP-1RA prescription and at least one weight measurement during therapy and within six months post-last prescription. Patients with prior prescriptions for other GLP-1RAs were excluded to avoid confounding from treatment switching.
• Index Date: The index date was defined as the date of the last GLP-1RA prescription followed by at least six months of no further prescriptions for that medication.
• Weight Trajectory Analysis: Weight changes were calculated as percentage differences from baseline weight. Data was aggregated into 30-day bins from 12 months pre-index to 6 months post-index. Patients were categorized into two groups: "weight regain" (≥2% weight increase in the six months post-last prescription) and "non-regain" (<2% increase).
• Clinical Trial-Style Proxy: A secondary, more stringent analysis restricted the cohort to patients with no subsequent GLP-1RA prescription for one year post-index to mimic clinical trial discontinuation protocols.
• Lifestyle Factor Extraction: Using Large Language Models (LLMs) with schema-constrained decoding (GPT-OSS 20B), the authors analyzed longitudinal clinical notes from a representative subset (N=300) to extract frequencies of documented diet and exercise counseling.
• Statistical Analysis: Chi-square tests were used to evaluate differences in lifestyle counseling documentation between regain and non-regain groups.

Key Results

• Weight Trajectories: The distribution of weight change post-last prescription approximated a normal distribution with a mean of 0.29% (median 0). Contrary to the assumption of universal regain, approximately two-thirds of patients in both cohorts exhibited stable weight or continued weight loss in the six months following their last prescription.
  • Semaglutide Cohort (N=2,567): 33.0% experienced weight regain; 67.0% did not.
  • Tirzepatide Cohort (N=1,615): 27.6% experienced weight regain; 72.4% did not.
• Clinical Trial-Style Discontinuation: In the subset of patients with no GLP-1RA prescription for one year post-index, weight regain remained a minority outcome: 39.3% for semaglutide and 26.6% for tirzepatide.
• Lifestyle Factors: In the representative subset (N=300), patients who did not regain weight were significantly more likely to have documented exercise counseling compared to those who regained weight (26.2% vs. 14.7%; p=0.04). Rates of diet counseling were similar between groups (22.3% vs. 23.5%; p=0.83).
• Discontinuation Documentation: Clinician documentation of treatment cessation within ±90 days of the last prescription was found in 39.5% of semaglutide regain patients and 44.0% of non-regain patients. Documentation of outside-source GLP-1RA use was rare in the notes (1.7% and 1.8% in non-regain groups), though the authors note this likely underestimates true prevalence due to reporting limitations.

Significance and Claims
The authors claim that this study provides the first real-world analysis of weight trajectories anchored specifically to the last GLP-1RA prescription by a physician. The findings challenge the notion that weight regain is an inevitable consequence of stopping therapy. Instead, the data suggests that a majority of patients maintain or continue to lose weight after the last prescription, potentially due to a combination of factors including sustained pharmacological effects, behavioral adaptations (specifically exercise), and continued access to medication from unrecorded sources.

The paper posits that these results do not contradict Randomized Controlled Trials (RCTs) but rather reflect the heterogeneity of real-world treatment cessation, which often involves partial discontinuation, staggered re-initiation, or lifestyle changes. The authors argue that these findings support a "longitudinal disease management paradigm" where clinicians can better identify patients who may safely discontinue therapy versus those requiring maintenance. They emphasize the need for future multimodal monitoring integrating biology, behavior, and dosing to predict which patients are likely to sustain benefits versus those at risk of rebound.

Limitations Acknowledged by the Authors

• Retrospective Nature: The study is subject to confounding variables inherent in real-world data.
• Follow-up Duration: The primary analysis is limited to six months post-prescription, which may underestimate long-term regain.
• Adherence Uncertainty: While inclusion criteria required three prescriptions, true adherence leading up to the last prescription cannot be guaranteed.
• Data Gaps: The study did not account for social determinants of health (e.g., insurance status) or comorbidities that impact access and adherence. Documentation of outside-source medication use is likely incomplete.