STEP 4 trial results showed that continuing semaglutide led to more weight loss, while stopping led to regain on average. Participants had already lost 10.6% during the first 20 weeks. Over the next 48 weeks, those continuing lost another 7.9%. Those switched to placebo, a look-alike injection without active medicine, gained 6.9%. Both groups kept receiving help with eating and physical activity.
What question did STEP 4 ask?
Once someone has lost weight on semaglutide, what happens if the medicine stops?
STEP 4 used randomized withdrawal: everyone started on the medicine, then some were switched by chance to placebo. That let researchers compare continuation with withdrawal while lifestyle support stayed in place.
Domenica Rubino and colleagues published the trial in JAMA in 2021. Its registration, NCT03548987, records the planned comparison. Novo Nordisk funded the study and participated in its design and analysis.
Who took part, and who reached the switch?
The study started 902 adults aged 18 or older. Participants needed a BMI, a measure of weight relative to height, of at least 30. A BMI of at least 27 qualified with a weight-related condition, such as high blood pressure. Everyone reported an unsuccessful dietary attempt to lose weight.
People with diabetes were excluded. So were people reporting a weight change greater than 5 kg, about 11 pounds, in the previous 90 days.
The run-in, the initial period when everyone received semaglutide, lasted 20 weeks. It included 16 weeks of gradually increasing doses and four weeks at 2.4 mg weekly. These were study doses; the prescriber sets an individual's dose.
Only 803 people reached the switch: 535 continued semaglutide and 268 received placebo. They averaged 46 years old; 79% were women and 83.7% were White. Neither participants nor investigators knew the assignment after the switch.
The comparison therefore describes people who reached and stayed on the target dose. It leaves out the 99 who didn't reach randomization.
What did the researchers measure?
The primary endpoint, the main measurement chosen beforehand, was percentage weight change from week 20 to week 68. Researchers also measured waist size, blood pressure and reported physical functioning, meaning how people managed everyday activities.
An estimand is the precise question an analysis answers. The main treatment policy estimand counted people in assigned groups, even after treatment stops or other weight-management help. Missing weights were estimated statistically.
A separate trial product estimand assumed everyone stayed on assigned treatment without other weight-management treatment. That analysis found −8.8% versus +6.5%. It answers a different question from the main −7.9% versus +6.9% comparison.
STEP 4 trial results: continuation versus withdrawal
The headline gap was 14.8 percentage points, the difference between a 7.9% decrease and a 6.9% increase over 48 weeks.
| Weight measurement | Continued semaglutide | Switched to placebo |
|---|---|---|
| Change from week 20 to week 68 | −7.9% | +6.9% |
| Change from the original start to week 68 | −17.4% | −5.0% |
| Gained weight after the switch, among people with measurements | 15.2% | 82.4% |
The first two rows are estimated averages from the main analysis. The last row uses observed measurements at week 68: 519 people continuing semaglutide and 250 receiving placebo. About four in five in the placebo group gained weight after the switch.
The 6.9% gain is measured from weight at week 20. It isn't the percentage of lost weight that returned. The placebo group still finished an average of 5.0% below its original starting weight. These percentages use different starting points, so adding them together gives the wrong answer.
The paper's results and tables also show waist size falling another 6.4 cm with continuation versus rising 3.3 cm with placebo. Systolic blood pressure, the top number, changed by +0.5 versus +4.4 points.
What side effects were recorded?
Stomach and bowel problems were common during the first 20 weeks: 71.4% reported them, and 5.3% stopped treatment because of adverse events. An adverse event is a health problem recorded during a study, whether or not the medicine caused it.
These selected rows come from Table 3, after randomization. They describe 535 people continuing semaglutide and 268 switched to placebo.
| Recorded event or treatment stop | Continued semaglutide | Placebo |
|---|---|---|
| Diarrhea | 14.4% | 7.1% |
| Nausea | 14.0% | 4.9% |
| Constipation | 11.6% | 6.3% |
| Vomiting | 10.3% | 3.0% |
| Serious events, such as those requiring hospitalization | 7.7% | 5.6% |
| Stopped treatment because of adverse events | 2.4% | 2.2% |
Most digestive events were mild to moderate. The low treatment-stop rate after week 20 needs its earlier context: people unable to continue had already left the comparison.
Where the findings need care
The authors' own limitations section explains that the run-in selected people who tolerated the schedule and stayed with treatment. That can make results look better than they would among everyone starting semaglutide in ordinary care. Researchers also didn't measure how closely participants followed the lifestyle plan.
STEP 4 had no randomized tapering arm or planned comparison of lower maintenance doses. Some participants could receive a reduced dose for tolerability, but that doesn't establish a lower-dose strategy.
The main weight comparison ended at week 68, only 48 weeks after the switch. Safety follow-up continued for another seven weeks. The study cannot show where weight regain eventually levels off or whether every participant would return to starting weight.
What later studies added
The STEP 1 extension, published in Diabetes, Obesity and Metabolism in 2022, followed a smaller group for a year after treatment ended. Unlike STEP 4, medication and organized lifestyle support stopped together. The STEP 1 trial explainer covers that follow-up and its limits.
SURMOUNT-4, the corresponding tirzepatide withdrawal trial, also found regain after switching to placebo; it cannot rank the two medicines against each other.
As of September 30, 2026, the day this article was written, a new analysis published August 31, 2026 in Endocrinology, Diabetes & Metabolism had brought STEP 4 together with other withdrawal studies. It reanalyzed six studies involving 1,776 people; its semaglutide subgroup included 634 people across three studies. The model estimated average regain of 0.89 kg, about 2 pounds, per month for that subgroup. Observations ended within 52 weeks, so longer-term estimates were projections that assumed a steady regain rate, rather than measurements of what happened later.
What this means for a prescription
STEP 4 supports discussing maintenance early. It doesn't assign anyone a lifelong schedule. The broader guide to what happens when semaglutide stops covers the practical care questions.
Through Promise, semaglutide is dispensed as a compounded medication, prepared by a licensed pharmacy for an individual prescription, which is different from an FDA-approved product: the formulation offered here is not FDA-approved. The FDA's compounding explanation, marked “Content current as of: 09/16/2025,” says it doesn't verify these preparations' safety, effectiveness or quality before marketing. STEP 4 did not test Promise's formulation, so its numbers don't establish that preparation's results.
A licensed provider may still prescribe a compounded formulation; that decision is between the patient and the doctor. At Promise, a licensed provider reviews every request, and not everyone qualifies. Whether to continue, reduce or stop treatment depends on the person's health history, response and manageable side effects.
