Of 58,618 adults with at least 5% initial weight loss and evaluable one-year trajectories, 8,389 (14.3%) regained at least five percentage points from their nadir. Regain was associated with inflammatory and acute-phase changes and higher cardiovascular and renal event rates, but the EHR could not verify drug ingestion or establish that regain caused those events.
The research question
How often does substantial weight regain occur after initial semaglutide or tirzepatide response, including when treatment remains documented, and what clinical signals accompany it?
Overview
Incretin withdrawal trials show that weight often returns after therapy stops, but routine-care rebound may also occur while treatment appears to continue. Investigators aligned each patient’s clinical history to their own lowest recorded weight, or nadir, to characterize documented treatment, laboratory changes, gastrointestinal and infectious diagnoses, and later clinical events surrounding weight regain after semaglutide or tirzepatide initiation.
Study design & methods
This retrospective cohort used de-identified EHR data from a federated U.S. network of more than 29 million patients. Adults initiating semaglutide or tirzepatide from 2021 through 2024 were required to have pre-index encounters and weights, no other incretin use in the prior year, no bariatric surgery, a further index-drug order within one year, and sufficient post-index weights. Among 58,618 patients who achieved at least 5% weight loss and had an evaluable one-year trajectory, regain was defined as at least five percentage points gained from the first-year nadir; 8,389 met that definition and 50,229 were sustained responders. Treatment documentation combined structured orders with NLP-extracted affirmative note mentions carried forward for 90 days. Laboratory and diagnosis patterns were aligned to the nadir, and time-varying event rates were adjusted for follow-up interval, brand, and age with Benjamini–Hochberg correction.
Regained weight tracked inflammatory changes and higher cardiovascular event rates
Weight regain occurred in 8,389 of 58,618 patients (14.3%). Among regain patients with at least 90 post-nadir days, 45.8% remained documented on treatment for at least half of follow-up at a dose maintained at or above their pre-nadir maximum, compared with 51.6% of sustained responders. At the nadir, regain patients had larger increases in neutrophil-to-lymphocyte ratio (+0.73 versus +0.29) and neutrophils (+0.48 versus +0.12 ×10⁹/L), lower albumin change (−0.12 versus −0.03 g/dL), and falling rather than rising HDL (−0.65 versus +1.03 mg/dL; all P less than 0.001). During regained versus maintained-loss time, adjusted rates were higher for nonfatal major adverse cardiovascular events (12.9 versus 9.6 per 1,000 person-years; RR 1.43, 95% CI 1.22–1.67), heart-failure hospitalization (20.1 versus 13.6; RR 1.66, 1.45–1.89), and cardiac arrest (1.8 versus 0.9; RR 2.11, 1.37–3.26; all q less than 0.001).
Limitations & interpretation
These results support weight nadir as a potentially useful clinical transition point but do not show that continued medication exposure failed or that regain caused later cardiovascular events. Prescriptions and note mentions do not confirm dispensing or ingestion, and blinded physician review confirmed ongoing use in only 17 of 25 adjudicable patients categorized as maintained treatment. The regain and sustained-response groups differed at baseline; illness, medication pauses, reverse causation, visit frequency, selective laboratory testing, incomplete external prescribing data, and the first-year nadir definition could influence the findings. Generalizability is limited to patients with sufficiently dense EHR follow-up. Prospective studies with objective exposure measurement are needed. Evidence confidence: Moderate. Rationale: a large longitudinal cohort and consistent signals support the associations, but treatment exposure, residual confounding, and reverse causation remain unresolved. Confidence variables: design—retrospective observational; sample size—large; exposure—documented, not verified; generalizability—care-engaged U.S. patients.
Publication type and venue follow the original nference listing. Read the original paper for full methods, analyses, and interpretation. View source listing.