Maintenance and stopping: what STEP 4, SURMOUNT-4 and the STEP 1 extension measured

Three trials answer the question most people eventually ask. Two are randomised withdrawal trials, the strongest design for it. One is an observational follow-up. Here is what each measured.

STEP 4: semaglutide

Design: 902 adults started semaglutide and escalated to 2.4 mg over 20 weeks. The 803 who reached the dose were randomised to continue or to switch to placebo for 48 more weeks, double-blind. Diet and activity counselling continued for everyone.

Result: mean loss over the 20-week run-in was 10.6 percent. From week 20 to 68, the continued group lost a further 7.9 percent of their week-20 weight; the placebo group regained 6.9 percent. Difference -14.8 percentage points (95% CI -16.0 to -13.5). Net from baseline at week 68: about -17.7 percent continued, about -4.4 percent stopped (PubMed 33755728).

SURMOUNT-4: tirzepatide

Design: 783 adults took open-label tirzepatide, escalated to 10 or 15 mg, for 36 weeks. The 670 who continued were randomised to tirzepatide or placebo for 52 weeks, double-blind, with counselling throughout.

Result: mean loss over the lead-in was 20.9 percent. From week 36 to 88, the continued group lost a further 5.5 percent of week-36 weight; the placebo group regained 14.0 percent. Difference -19.4 points (95% CI -21.2 to -17.7). Net from baseline at week 88: about -25.3 percent continued, about -9.9 percent stopped (PubMed 38078870).

Normalised to baseline, the SURMOUNT-4 placebo arm regained about 53 percent of what it had lost in a year; the STEP 4 placebo arm about 58 percent. The maintenance planner shows that arithmetic and applies it to your weight.

STEP 1 extension: one year off semaglutide

After STEP 1 ended at week 68, 327 participants were followed for a further 52 weeks with no drug and no counselling. The semaglutide group had reached -17.3 percent at week 68; at week 120 they were at -5.6 percent, having regained 11.6 percentage points, about two-thirds of the loss. Cardiometabolic improvements (blood pressure, lipids, HbA1c) reverted towards baseline alongside (PubMed 35441470). This is observational (no randomised comparison), but the direction matches the two randomised trials.

What the trials do not tell you

• Your course. The spread inside each placebo arm was wide. A mean of +14 percent contains people who regained everything and people who regained little.
• Tapering. All three trials stopped abruptly. No published phase 3 trial has compared a taper with a stop.
• Lower doses for maintenance. Not tested in these trials. A prescriber may have views; the trials do not.
• Compounded products. Not FDA approved, not interchangeable with brand products, and not what was studied.
• Who keeps the loss. The trials did not identify predictors reliable enough to use. Regain was the rule on average.

What has evidence for holding a loss

The S-LiTE trial randomised 195 adults who had lost 13.1 kg on an 8-week low-calorie diet to placebo, supervised exercise, liraglutide 3.0 mg, or both, for a year. Exercise alone held 4.1 kg more than placebo, liraglutide alone 6.8 kg more, and the combination 9.5 kg more; the combination roughly doubled the improvement in body-fat percentage seen with either alone (PubMed 33951361). It is liraglutide, not semaglutide or tirzepatide, and the exercise was supervised and substantial (about 150 minutes a week of moderate-to-vigorous activity), but it is the cleanest evidence that structured exercise adds to and partly substitutes for drug effect on maintenance.

Beyond that: the withdrawal trials continued diet and activity counselling in every arm, and every arm off drug regained. So counselling as delivered in trials did not prevent regain. Protein intake and resistance training protect what you keep as lean tissue (see resistance training and protein); weekly weighing detects regain early enough to act (see how to weigh yourself).

Turning this into a conversation

The decision to continue, stop or change is a prescriber's, informed by you. The planner builds a question list from what is driving your decision: reaching a goal, side effects, cost, supply, pregnancy planning, surgery, other medications. It contains no dosing suggestions. Take the list, your weight log and your reasons to the appointment. If cost is the driver, the FormBlends GLP-1 cost report lays out the current price landscape, and the FormBlends GLP-1 program describes what FormBlends itself offers, with the disclosure that FormBlends dispenses compounded products that are not FDA approved.