What Tirzepatide Maintenance Trials Can—and Cannot—Promise About Keeping Weight Off

01SURMOUNT-4 randomized a decision after treatment had already worked

SURMOUNT-4 was a phase 3 randomized withdrawal trial at 70 sites in four countries. Eligible adults had a body mass index, or BMI, of at least 30, or at least 27 with a weight-related complication. Diabetes was excluded. Everyone entering the initial phase received once-weekly subcutaneous tirzepatide, meaning an injection under the skin, with diet and physical activity. The studied maximum tolerated doses were 10 or 15 mg; these identify trial arms, not instructions for treatment.

The first 36 weeks were open-label: participants and investigators knew tirzepatide was being used. Only then were 670 participants assigned equally to continue tirzepatide or switch to placebo for 52 weeks. This second phase was double-blind, with treatment assignment concealed. Randomizing at that point creates a direct test of continuation versus withdrawal among people who reached the maintenance decision.

That is different from asking what happens to everyone starting a prescription. The randomized participants had a mean age of 48 years, 71% were women, and the report gives a mean weight of 107.3 kg. The trial is most informative about a selected, treatment-experienced population, rather than all prospective service customers.

02Why the difference between 783 and 670 matters

Of 783 adults entering the lead-in, 670 reached randomization: about 86%, leaving 113 outside the randomized comparison. The indexed abstract does not give a complete breakdown of why those individuals did not proceed. It would therefore be incorrect to label all 113 as adverse-effect withdrawals, treatment failures, or administrative losses.

The important design consequence is selection. Maintenance participants had completed a substantial period of treatment and reached a maximum tolerated study dose. They were selected tolerators and completers, not an unfiltered sample of everyone who might try tirzepatide. Their 20.9% average lead-in loss describes this randomized population, not a guaranteed starting-treatment outcome for all 783 entrants.

Selection does not erase the randomized result. Within the group that reached week 36, random assignment supports a strong causal comparison between continuing treatment and switching to placebo. It does narrow the question being answered. A service claim about every new enrollee needs evidence that includes early discontinuation, inadequate response, and inability to tolerate treatment—not just outcomes among people who successfully reach maintenance.

03The percentage arithmetic behind loss and regain

The primary endpoint was mean percentage weight change from week 36 to week 88. Continued tirzepatide produced a further 5.5% reduction, while placebo produced a 14.0% increase. The reported between-group difference was −19.4 percentage points, with a 95% confidence interval from −21.2 to −17.7. This is a difference in weight-change percentages, not a relative reduction in the risk of gaining weight.

A worked illustration clarifies the denominator. Imagine an initial weight of 100 arbitrary units. A 20.9% loss leaves 79.1 units at week 36. A subsequent 14.0% increase is calculated from 79.1, adding about 11.1 units—not 14 units of the original weight. A further 5.5% decrease similarly subtracts about 4.4 units. This is illustrative arithmetic, not an individual forecast or a reconstruction of the statistical model.

The reported overall losses from week 0 to week 88 were 25.3% with continued tirzepatide and 9.9% after switching to placebo. Thus, substantial regain did not mean that the placebo group, on average, exceeded its original weight. Nor can the rounded phase-specific percentages simply be added or subtracted as though they share one denominator.

04Maintaining 80% of a loss is not the same as never regaining

The key maintenance endpoint counted participants who retained at least 80% of the weight they had lost during the lead-in. For someone who initially lost 20 weight units, that threshold would mean retaining at least 16 units of loss. Some regain could therefore occur while the person still met the study’s maintenance definition.

At week 88, 300 participants continuing tirzepatide, or 89.5%, met that threshold, compared with 16.6% receiving placebo. The absolute difference was 72.9 percentage points. This categorical result complements the average weight changes: it shows that the continuation advantage was not expressed only as a difference between two group means.

It still does not establish that every continuing participant maintained all their loss, or that every participant withdrawing treatment regained it. The confidence interval for the primary mean difference describes uncertainty around the group comparison, not the range of individual outcomes. Readers comparing withdrawal evidence can find the separate STEP 1 extension analysis, whose endpoint and design should not be treated as interchangeable with this randomized maintenance comparison.

05SURMOUNT-MAINTAIN tested an additional maintenance strategy

It is no longer accurate to say that maintenance-dose reduction has not been studied. SURMOUNT-MAINTAIN was a 112-week phase 3b trial at 20 US sites. It enrolled 441 adults into a 60-week open-label weight-loss period using weekly subcutaneous tirzepatide at a maximum tolerated dose of 10 or 15 mg. At week 60, 378 participants were randomized: 140 continued that dose, 144 entered a 5 mg arm, and 94 switched to placebo.

Eligibility included BMI of at least 30, or at least 27 with a weight-related condition, and at least one self-reported unsuccessful dietary weight-loss effort. The enrolled population had a mean age of 46.6 years, mean weight of 113.8 kg, and mean BMI of 40.1; 65% were female and 67% were White. These characteristics help define the population behind the averages.

After another 52 weeks, model-based weight changes from the original baseline were −21.9% with continued maximum tolerated treatment, −16.6% in the 5 mg arm, and −9.9% with placebo. Those are total 112-week changes, not additional losses after randomization. The pattern supports a lower-dose strategy as an evidence-based research finding, while showing less retained average weight reduction than continued maximum tolerated treatment.

06Rescue treatment changes how the newer results should be read

Beginning at week 84, 24 weeks after randomization, SURMOUNT-MAINTAIN allowed rescue tirzepatide when regain exceeded 50% of previously lost weight. The observed rescue-treatment proportions were 8%, 25%, and 67% in the maximum tolerated, 5 mg, and placebo arms, respectively. The denominators in that report differ from the original randomized group sizes, so those figures should retain their stated denominators.

The primary analysis used a modified treatment-regimen estimand. An estimand is the precise treatment effect a statistical analysis aims to measure, including how it handles events such as treatment discontinuation or rescue. Here, all randomized participants were included regardless of discontinuation or prohibited medication use, with an assumption that those initiating rescue would gain no further benefit from their assigned study treatment. The results are therefore not simply untreated observations after rescue.

Compared with placebo, estimated differences were −12.0 percentage points for maximum tolerated treatment and −6.6 for the 5 mg arm. The latter had a 95% confidence interval of −8.3 to −5.0; the former interval was approximately −13.8 to −10.1, although its indexed rendering contains a typographical artifact. Overall study completion was 345 of 378, or 91%. Eli Lilly funded this trial.

07Safety findings must be read alongside the selected population

In SURMOUNT-4, the most common adverse events were gastrointestinal, mostly mild to moderate, and occurred more often with tirzepatide than placebo. The indexed abstract does not provide event-specific rates or a detailed serious-adverse-event comparison. It supports reporting a real tolerability burden, but not calculating a precise excess rate for a particular symptom.

SURMOUNT-MAINTAIN likewise reported gastrointestinal events as the most common adverse events with tirzepatide. They were mostly mild to moderate and occurred mainly during dose escalation. Its maintenance safety population included 372 participants receiving at least one maintenance-period dose: 139, 142, and 91 across the three arms. Neither abstract supports a blanket assertion that maintenance is free of important risks.

Both trials began maintenance comparisons after lengthy exposure, which matters when interpreting tolerability for new starters. The evidence also does not identify the best strategy for every individual, establish a tapering schedule, or compare all possible formulations. A direct-comparison trial addresses a different question from withdrawal; treatment rankings elsewhere cannot supply missing maintenance safety data here.

08Semaglutide withdrawal provides context, not a head-to-head verdict

The STEP 1 extension followed 327 participants from a much larger semaglutide trial. The original trial randomized 1,961 adults without diabetes to weekly subcutaneous semaglutide 2.4 mg or placebo for 68 weeks. At treatment completion, medication and structured lifestyle intervention stopped. The extension then observed a selected subset for another year; its analyses were exploratory, rather than a newly randomized continuation-versus-withdrawal experiment.

In that subset, mean weight loss during treatment was 17.3% with semaglutide and 2.0% with placebo. By week 120, the former semaglutide group had regained 11.6 percentage points relative to original baseline, leaving a reported net loss of 5.6%. The standard deviations—9.3% for initial loss and 7.7 percentage points for regain—show substantial variability. The familiar description of two-thirds regained summarizes a group average, not everyone’s trajectory.

Most cardiometabolic measures moved back toward baseline, but not every improvement disappeared. Systolic blood pressure averaged 129, 121, and 131 mm Hg at baseline, week 68, and week 120, while some lipid and inflammatory measures retained improvement. These are risk-factor observations, not proof of changes in heart attacks or survival. The extension, funded by Novo Nordisk, did not systematically collect adverse events and cannot isolate medication withdrawal from withdrawal of structured lifestyle support.

09A provider’s maintenance claim needs its own documentation

A defensible clinical statement is that continued tirzepatide outperformed withdrawal for maintaining weight reduction in the studied populations. A stronger statement—that a particular service guarantees comparable maintenance—requires service-specific evidence. Documentation should identify the product, route, population, follow-up period, maintenance definition, and handling of people who leave care. Excluding unsuccessful participants can make commercial outcome summaries look more favorable than an all-enrollee analysis.

Editorial interpretation: a clinical maintenance benefit does not establish that a long prepaid commitment is medically necessary, that coverage will continue, or that a refund promise is easy to use. Those are separate documentary questions. The billing commitments guide and refund boundaries guide explain how to distinguish duration, eligibility conditions, and financial remedies from treatment outcomes.

Readers considering a named service can use the Mochi Health review to check documentary terms, not as evidence that the provider supplies the studied product or achieves these results. A legitimate guarantee should state exactly what is guaranteed—money, access, or a measured outcome—and what happens when treatment is unsuitable or interrupted. Randomization in a drug trial validates none of those contractual promises.

10Product identity and follow-up set the evidence boundary

The studied injection regimens cannot automatically validate compounded preparations, oral routes, other formulations, or a platform’s care model. FDA states that compounded drugs are not FDA approved and are not reviewed before marketing for safety, effectiveness, or quality. It recommends compounded products only when a patient’s needs cannot be met by an FDA-approved drug. Its concerns include dosing errors and semaglutide salt ingredients; adverse-event reports alone do not establish incidence or causation.

The practical documentary question is therefore what product a service actually proposes, not merely whether its marketing mentions tirzepatide research. The product terms guide separates formulation descriptions from evidence claims. This distinction does not require assuming that every preparation has the same problem; it requires recognizing that equivalence has not been established by these maintenance trials.

Important scientific questions remain: which participants can sustain benefit with reduced treatment, how outcomes develop beyond these maintenance windows, and what happens with different support arrangements or interruptions. NIDDK describes duration as depending on benefit and tolerability, with cost and other medicines also relevant. That supports individualized clinical reassessment, not a universal stopping date or a claim that every participant must remain on one unchanged regimen indefinitely.

Questions & answers

Sources & documentation

Scientific and official sources consulted October 20, 2026. Study findings apply to the populations, products and time periods described; they are not provider outcome data.

1. Continued Treatment With Tirzepatide for Maintenance of Weight Reduction in Adults With Obesity: The SURMOUNT-4 Randomized Clinical Trial. ↗
2. Weight regain and cardiometabolic effects after withdrawal of semaglutide: The STEP 1 trial extension. ↗
3. Tirzepatide for maintenance of bodyweight reduction in people with obesity in the USA (SURMOUNT-MAINTAIN): a multicentre, double-blind, randomised, placebo-controlled trial. ↗
4. NIDDK prescription medications for overweight and obesity ↗
5. FDA concerns about unapproved GLP-1 drugs ↗

Continue reading

• What STEP 1 Weight Regain Can—and Cannot—Support in GLP-1 Service Claims
• What SURMOUNT-5 Proves—and What GLP-1 Service Claims Cannot Borrow
• Why Clear GLP-1 Contracts Do Not Prove Clinical Outcomes
• Approved, Compounded and Other GLP-1 Formulations: Read the Product Before Signing
• A Monthly Price Is Not Always a Monthly Commitment
• GLP-1 Refund Guarantees: What Is Covered, What Is Excluded, and When Money Returns