A meta-analysis published in Endocrinology, Diabetes & Metabolism in September 2026 puts a number on something patients ask about far more often than they are answered: what happens after you stop. Pooling six studies and 1,776 participants, it estimates weight comes back at 1.04 kg per month, that half of everything lost is back within about seven and a half months, and that starting weight is reached again at roughly fifteen months. The last of those three figures is an extrapolation, and that distinction matters more than the number.
What the analysis did
The authors did not run a new trial. They took published timepoint-level aggregate data from six studies — ten treatment arms in all, with observed follow-up ranging from 4 to 52 weeks after discontinuation — and re-analysed them in a Bayesian hierarchical longitudinal model. The model estimated four things: how much weight people had lost when they stopped, how fast they regained it, when half the loss would be gone, and when they would be back where they started.
Weight loss at cessation was 15.35 kg (95% credible interval 11.18–19.60). Monthly regain was 1.04 kg (0.80–1.29), which works out to roughly 12.5 kg a year if the rate holds. Half the initial loss was projected to return by 7.50 months (5.05–10.57) and all of it by 15.00 months (10.10–21.13) — and you can check that arithmetic yourself: 15.35 kg divided by 1.04 kg per month is 14.8 months, which is simply the straight line the model drew, not an observation anyone made. Tirzepatide showed numerically faster regain than semaglutide before adjustment, but no clear drug-specific difference survived adjustment. Greater initial weight loss had the strongest directional association with faster regain, and post-discontinuation behavioural or lifestyle support was directionally associated with slower regain — but the credible interval for the first included zero and the second was too imprecise to act on.
What it changes, and what it does not
It moves the discontinuation question from anecdote to an estimate with an interval around it, and the estimate is unflattering. A rate of about a kilogram a month means the difference between stopping and not stopping shows up on a bathroom scale within weeks, not years. That has a practical consequence for how these drugs are priced and planned for: the arithmetic that makes a course of treatment look affordable over six months looks very different if the alternative to continuing is losing most of the benefit inside a year. Readers weighing what these drugs cost without insurance should treat the monthly figure as an ongoing cost rather than a course price. It also argues for settling the question of what happens next before stopping rather than after, which is a conversation to have with whoever is prescribing, not one to improvise.
This is a re-analysis of aggregate published data, not individual patient data, so it inherits every limitation of the six source studies and can adjust for very little. Observed follow-up went no further than 52 weeks; the return-to-baseline estimate assumes the regain rate stays constant, and the authors explicitly warn that anything beyond a year is a model-based extrapolation. Real regain curves usually flatten rather than continue straight, so fifteen months may well be pessimistic. Six studies is a thin base for meta-regression, which is why the predictors of faster regain come back directional but inconclusive. And the analysis says nothing about who these people were — whether they stopped by choice, by cost, or by side effect — which is likely to matter for what they did next.
Sources: Kow CS, et al. “Weight Regain Trajectories After Discontinuation of Semaglutide or Tirzepatide: A Reconstructed Aggregate-Data Bayesian Longitudinal Meta-Analysis.” Endocrinology, Diabetes & Metabolism, September 2026;9(5):e70325. PubMed 42673571 · doi:10.1002/edm2.70325
