Most of the research on GLP-1 medications asks what happens while people take them. A growing share asks the harder question: what happens after they stop. We have written before about the pattern, which is that much of the weight tends to come back over time.
A new study takes the next step. Instead of measuring regain, it tried to slow it. The result is encouraging, and it comes from a trial small enough that the encouraging part needs to be read carefully.
What the trial did
The study, published October 7 in Obesity Science & Practice by researchers at Tufts University’s Food is Medicine Institute, was a randomized pilot trial run remotely between March 2024 and August 2025.
It enrolled 39 adults who had lost more than 10% of their body weight on a GLP-1 receptor agonist and had stopped the medication within the previous 30 days. On average they had lost about 18% of their starting weight while on treatment. Participants were randomly assigned to one of three groups for four months:
- Medically tailored meals (13 people): ten meals a week, chosen from a menu of about 70 and delivered to their homes.
- A digital wellness app (12 people): a Noom subscription, with a request to use it at least 15 minutes a day.
- Usual care (14 people): whatever their own clinician recommended, with no study intervention.
The meals and the app were provided free. Weight was measured at home on Bluetooth scales, with monthly weigh-ins requested.
What it found
Over four months, the model estimated that the usual-care group regained about 10.75% of their body weight. The app group regained about 3.60%, and the meals group about 3.58%. The difference held after the researchers adjusted for how much weight each person had lost on the medication beforehand.
Two things stand out. Both structured options did meaningfully better than usual care in this sample. And the two options did almost exactly as well as each other, which hints that the specific tool may matter less than having some structure in place during the transition off the medication.
Corresponding author Kelseanna Hollis-Hansen framed the first year after stopping as the period where support matters most, while noting that how long it needs to last is still an open question.
What the trial does not establish
This is a genuinely useful study. It is also a pilot, and a pilot is designed to test whether a bigger trial is feasible, not to settle the question.
Weight was not the main thing it measured. The trial’s primary endpoints were adherence, engagement, and satisfaction, and none of those differed significantly between groups. Weight change was an exploratory outcome. Exploratory results are how researchers decide what to test properly next; they are not designed to stand on their own.
Twelve to fourteen people per group is very small. With groups this size, a couple of participants with unusual trajectories can move an average a long way. The study also had missing weight measurements, and adherence was self-reported.
Four months is short. The authors say plainly that the trial cannot show lasting weight maintenance or any heart or metabolic benefit. Whether the gap between groups widens, holds, or closes after the meals stop arriving is unknown.
Both groups still regained weight. The interventions slowed regain in this sample. They did not prevent it.
The participants were not a broad cross-section. About 82% were women, most were college-educated and employed, and recruitment came through Texas obesity clinics and a research-matching platform. People with certain psychiatric or medical conditions were excluded.
Free is doing some work here. Ten delivered meals a week and an app subscription cost real money outside a study. A result achieved when both are free does not tell us how many people would sustain them when paying out of pocket.
It did not test any medication. The trial compared meals, an app, and usual care after people had already stopped. It says nothing about tapering, lower maintenance doses, or restarting, and it is not evidence that anyone should stop treatment.
Where compounded medications fit
The coverage we could access does not name which GLP-1 medications participants had been taking. More broadly, the trials that established how much weight people lose on these medications, and how much tends to return after stopping, studied branded, manufactured products.
Compounded semaglutide and tirzepatide, which NoBsRx provides provider-guided access to, are prepared by state-licensed pharmacies for individual patients. Compounded medications are not FDA-approved, and they have not been studied in trials like these. Evidence from a branded product, including what happens when someone stops it, does not automatically transfer to a compounded preparation.
Why this matters anyway
The practical takeaway does not depend on which product someone used. Stopping a GLP-1 is a transition, not just an absence, and this trial is early evidence that planning for that transition may be worth the effort.
That is part of why we think the end of treatment belongs in the conversation from the start, not as an afterthought. Decisions about when, whether, and how to stop are clinical ones, made with the licensed provider who knows your history, and answers to common questions about that process live in our FAQ. Meal structure, coaching, and follow-up are all reasonable things to bring into that conversation, with the honest caveat that the evidence for any specific plan is still thin.
A larger, longer trial is the obvious next step. We will be watching for it.
This post is general information, not medical advice. It does not describe a benefit of any NoBsRx program and it is not a claim that any program, meal plan, or app will maintain weight loss. Compounded medications are not FDA-approved. Treatment is never guaranteed, and individual results vary. Eligibility and every treatment decision, including whether to start, continue, or stop a medication, are made by independent licensed providers based on an individual clinical assessment, and availability varies by state. Do not change any treatment based on an article.