Use was increasing rapidly. Compounded users had lower observed weight loss than trial participants, while heart-rate and symptom findings support closer surveillance but do not establish cardiovascular harm.
The research question
What did clinical records show about unapproved retatrutide use, effectiveness, and early safety signals?
Overview
Clinical-note surveillance found rapidly growing use of products purported to contain retatrutide outside clinical trials. Compounded users lost 7.2% at six to twelve months versus 15.5% in trial participants, while early cardiovascular findings remain preliminary. An unapproved product leaves no ordinary prescription or claims trail. The study shows how evidence-linked review of clinical notes can identify emerging exposure and outcomes before structured surveillance exists. An unapproved product leaves no ordinary prescription or claims trail. The study shows how evidence-linked review of clinical notes can identify emerging exposure and outcomes before structured surveillance exists.
Study design & methods
Across a federated U.S. EHR network of 29 million patients, an LLM adjudicated full clinical notes for 983 people with a retatrutide mention. Exposure was confirmed for 652 and supply route for 531. Trial participants anchored 1:3:10:10 nearest-neighbor matching to compounded-retatrutide, semaglutide, and tirzepatide users on seven baseline covariates, producing groups of 89, 243, 890, and 890 patients. Across a federated U.S. EHR network of 29 million patients, an LLM adjudicated full clinical notes for 983 people with a retatrutide mention. Exposure was confirmed for 652 and supply route for 531. Trial participants anchored 1:3:10:10 nearest-neighbor matching to compounded-retatrutide, semaglutide, and tirzepatide users on seven baseline covariates, producing groups of 89, 243, 890, and 890 patients.
Principal findings
Among users with a known route, 71.2% obtained purported retatrutide outside a trial; use grew 1.8-fold per quarter. At six to twelve months, observed mean weight loss was 15.5% in trial participants and 7.2% among compounded users, similar to 7.7% with tirzepatide. Heart rate rose by 4.3 and 2.5 bpm at three months in the two retatrutide cohorts. Pooled retatrutide users had higher documented cardiovascular and neuropsychiatric symptom incidence than approved comparators, while MACE estimates were imprecise and not statistically significant. Among users with a known route, 71.2% obtained purported retatrutide outside a trial; use grew 1.8-fold per quarter. At six to twelve months, observed mean weight loss was 15.5% in trial participants and 7.2% among compounded users, similar to 7.7% with tirzepatide. Heart rate rose by 4.3 and 2.5 bpm at three months in the two retatrutide cohorts. Pooled retatrutide users had higher documented cardiovascular and neuropsychiatric symptom incidence than approved comparators, while MACE estimates were imprecise and not statistically significant.
Limitations & interpretation
The comparison suggests that gray-market products may deliver substantially less benefit while preserving biologically active exposure, but it cannot determine product contents, dose, adherence, or causality. Disclosure-based exposure, note-anchored start dates, residual imbalance, short uneven follow-up, surveillance bias, and small cardiovascular-event counts all limit interpretation. The paper is a preprint and should be read as an early pharmacovigilance signal, not proof of cardiovascular risk. The comparison suggests that gray-market products may deliver substantially less benefit while preserving biologically active exposure, but it cannot determine product contents, dose, adherence, or causality. Disclosure-based exposure, note-anchored start dates, residual imbalance, short uneven follow-up, surveillance bias, and small cardiovascular-event counts all limit interpretation. The paper is a preprint and should be read as an early pharmacovigilance signal, not proof of cardiovascular risk. Preprint; not peer reviewed.
Publication type and venue follow the original nference listing. Read the original paper for full methods, analyses, and interpretation. View source listing.