Most of Your Panel Will Pause, and Most of Those Will Come Back
At ENDO 2026 in June, Sainikhil Sontha of the Boston University School of Public Health presented an analysis of more than 60,000 Americans with type 2 diabetes. About 40% stopped their GLP-1 within the first year and nearly 60% had stopped by the end of two. That much matches what most clinics already feel.
The second half of the finding is the part almost nobody has built for. This is a conference presentation rather than a peer-reviewed paper, so treat the precision with some care, but it's hard to argue with the direction. As Sontha put it, use is "more start-and-stop than most people assumed."
The real-world evidence points the same way. A Cleveland Clinic study published in Diabetes, Obesity and Metabolism in March 2026 followed nearly 8,000 adults in Ohio and Florida who started semaglutide or tirzepatide and then stopped within three to twelve months. Of those, 27% switched to a different medication, 20% restarted the original, and 14% moved to lifestyle intervention. Only a minority simply stopped and did nothing.
So the thing your program treats as attrition is mostly a cycle. A patient who stops in March is, more often than not, a patient who's coming back.
Regain Is Front-Loaded, Which Tells You When to Act
Two different questions get asked about regain, and they produce very different numbers. Keeping them apart matters, because quoting the wrong one to a client is how a program loses an argument it should win.
The first question is what happens when a drug is withdrawn under trial conditions. In the STEP 1 trial extension, published in Diabetes, Obesity and Metabolism in 2022, roughly two-thirds of the loss came back, and most cardiometabolic gains reverted with it.
| Measure | Trial withdrawal (STEP 1) | Real-world stop (Cleveland Clinic) |
|---|---|---|
| Participants | 327 | Nearly 8,000 |
| Weight lost before stopping | 17.3% by week 68 | 8.4% |
| Change over the following year | Regained 11.6 percentage points | Regained 0.5% |
| Where they ended up | 5.6% below where they started | 45% maintaining or continuing to lose |
The second question is what happens to real patients, who switch and restart rather than stopping cold. That's the Cleveland Clinic cohort above. Note that these are two separate measures on two separate populations: a trial withdrawal and an observed real-world stop aren't the same event, and the gap between two-thirds and half a percent is mostly explained by what patients did next.
The most useful number sits between them. A systematic review and nonlinear meta-regression published in eClinicalMedicine in March 2026 pooled six randomized trials and 3,236 participants with weight data out to a year after cessation. Regain follows a decay curve rather than a straight line: it plateaus at an estimated 75.3% of the weight lost, with a half-life of twenty-three weeks, and about 40% of the on-treatment loss still remaining at one year.
Support delivered in month nine is arriving after the window that decided the outcome.
What GLP-1 Discontinuation Coaching Looks Like in Practice
The honest evidence here is narrower than the marketing around it, and it's worth knowing which claim you can defend. No trial has isolated what coaching adds on top of a GLP-1, because the trials combining them gave behavioral therapy to every arm.
What has been tested is whether adding structured support changes what happens after the drug stops. Jensen and colleagues, in eClinicalMedicine in 2024, randomized patients who had lost a mean 13.1 kg on diet to supervised exercise, liraglutide, both, or placebo for a year, then followed 109 of them through a second year with no treatment at all. Weight regain after stopping was 6.0 kg larger for the group coming off liraglutide alone than for the group coming off supervised exercise. Across the full two years, the group that had both finished 5.1 kg lighter than the drug-only group, with body fat 2.3 percentage points lower. Their conclusion is the operative sentence for a care team: adding supervised exercise to obesity pharmacotherapy "seems to improve healthy weight maintenance after treatment termination."
That points at four things to change.
Make the Stop a Trigger, Not a Silence
A discontinuation should fire a workflow the way a new enrollment does. Most programs only learn about it from a missed refill weeks later, which spends the window. It's the same failure as member drop-off between scheduled contacts, with higher stakes.
Front-Load the Cadence
If half the regain lands within six months, the contact schedule should be heaviest immediately after stopping and can taper after. Most programs have this exactly backwards, with the most attention at intake.
Protect Lean Mass, and Say Why
Across six semaglutide trials covering 1,541 adults, lean mass accounted for anywhere from almost none to 40% of total weight lost, published in Expert Opinion on Pharmacotherapy in 2024. Resistance training and adequate protein are sound physiology, though worth noting that the trials themselves didn't test structured exercise, so this is a clinical recommendation rather than a trial result. Our earlier piece on coaching GLP-1 patients covers that side at the individual level.
Treat a Restart as a Coaching Moment
Given that more than four in ten returners come back within the year, a restart is the most predictable event on your panel and the one most programs handle as a fresh intake. Building the panel around that cycle is what our work with GLP-1 and metabolic clinics is organized around.
Where Programs Get This Wrong
Two mistakes account for most of it. The first is measuring a GLP-1 program at twelve months against a continuous-treatment assumption, which produces a number that describes a population you don't have. The second is treating a discontinuation as a discharge, which hands the most consequential six months of the patient's year to nobody.
Neither is a coaching-quality problem. Both are design problems, and both are fixable without hiring anyone. Whether your current system can even detect a discontinuation is a separate question, and the criteria worth applying are in our guide to choosing coaching software.
