This is a Bayesian re-analysis of what happens after the drug stops. The authors pooled timepoint-level data from six discontinuation studies — ten treatment arms, 1,776 participants — and modelled the trajectory rather than reporting a single endpoint. Follow-up in the source studies ran from 4 to 52 weeks.
Weight loss at the point of stopping averaged 15.35 kg (95% credible interval 11.18 to 19.60). Regain then ran at 1.04 kg per month (0.80 to 1.29). On the model, half the weight lost is back by 7.50 months (5.05 to 10.57), and all of it by 15.00 months (10.10 to 21.13). Greater initial loss was directionally associated with faster regain, and behavioural support with slower regain, but both credible intervals included zero.
Why it matters: the rate is the useful part. A patient who asks what happens if they stop can now be told roughly a kilogram a month, which is concrete in a way that "most people regain the weight" is not. It also reframes these drugs as treatment for a chronic condition rather than a course with an end.
What to do about it: raise stopping before it happens, not after. Cost, supply and side effects all end prescriptions, and the month after is when the trajectory is set. Be careful with the fifteen-month figure — anything past 52 weeks here is a linear extrapolation, not something anybody observed, and real regain curves usually flatten.
- Quote the rate, not the outcome: about 1 kg a month after stopping
- Agree a maintenance plan before a course ends, whatever the reason it is ending
- Book a review within the first two months of stopping, when the slope is steepest
- Do not present the return-to-baseline figure as a prediction; it is a model extrapolation
- Ask what would happen to supply if cost or availability changed — that is the commonest reason people stop
The statistics, in plain English
A credible interval is the Bayesian counterpart of a confidence interval: 1.04 kg/month (0.80 to 1.29) means the data are most consistent with a kilogram a month, and would be surprised by less than 0.8 or more than 1.3. The wider intervals on the longer horizons — 15.00 months to baseline, but plausibly anywhere from 10 to 21 — are wide for a reason worth understanding. Observed follow-up stopped at 52 weeks, so everything past that is the model continuing a straight line. Weight regain rarely stays linear, so treat those as a warning about direction rather than a date to write down.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for diabetes & endocrinology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free