STEP Trials Summary: What Every Semaglutide Study Actually Found

By Weight Loss Provider Guide
Last verified: August 4, 2026 · STEP Evidence Atlas v1.0

There is no single STEP trial result — and that is the whole problem. This STEP trials summary covers 18 studies with peer-reviewed, FDA-label, or officially reported results. The famous number, 14.9%, comes from exactly one of them: STEP 1, in adults without type 2 diabetes, over 68 weeks. Change the people, the dose, the length, the support, or the statistical question and the number changes with it. Adults with type 2 diabetes averaged 9.6% in STEP 2. The current U.S. Wegovy label reports 18.8% for the newer 7.2 mg dose in STEP UP. Teenagers were measured primarily by change in BMI, so their result is not even the same kind of number.

The STEP program began as Novo Nordisk's phase 3 weight-management development program for semaglutide. Early trials supported Wegovy's original weight-management approval; later trials tested new populations, doses, conditions, and treatment questions. BARI-STEP belongs in this evidence atlas because it uses the STEP name and directly answers a post-bariatric question, but it was an investigator-led trial rather than one of Novo Nordisk's original registrational studies.

This page keeps the population, length, comparison group, endpoint, evidence status, and statistical method attached to every result — across STEP 1 through STEP 12, STEP TEENS, STEP UP, the heart-failure studies, and BARI-STEP.

One more thing before you scroll. The FDA's Wegovy prescribing information uses its own internal study numbers instead of the public STEP names. The two systems match in some places and break apart in others. It is easy to land on the wrong trial if you assume “Study 7” means STEP 7. We built the full crosswalk below.

Best for: figuring out which STEP study actually resembles your age, diabetes status, condition, dose question, and treatment question.

Not for: picking your own dose, predicting your personal result, or treating STEP findings as evidence for a compounded finished product.

Start here

What we actually verified

Checked on August 4, 2026:

• The current U.S. Wegovy prescribing information, revised 06/2026, read directly rather than summarized from a secondary source
• The primary paper, ClinicalTrials.gov record, or official congress report for every study in the tables below
• Population, number randomized, diabetes status, dose, comparison group, length, endpoint, analysis convention, headline result, and evidence status for each study
• The FDA's March 19, 2026 approval of Wegovy 7.2 mg and the current label instructions that govern that dose
• STEP Young and STEP TEENS Weight Maintenance registry status and whether results had been posted
• Ro's current membership plans, free insurance-checker scope, dose-specific Wegovy prices, and the published expiration dates for the two introductory Wegovy offers

We did not reanalyze patient-level data, turn a trial average into a personal prediction, or treat conference-reported findings as peer-reviewed findings. Where evidence is provisional, specialized, or not directly comparable with STEP 1, the row says so.

STEP trials summary: what did all the studies find?

Answer: Across the completed randomized studies in this atlas, results on the prespecified primary question favored semaglutide over the comparison group, but there is no single STEP weight-loss percentage. Main adult body-weight results ranged from 9.6% in STEP 2 to 18.8% in the current U.S. label for STEP UP, while withdrawal, pediatric, head-to-head, heart-failure, osteoarthritis, prediabetes, and post-bariatric studies answer different questions and should not be ranked as though they were interchangeable.

Here is the program in one normalized table. The extra columns are the point: evidence status tells you how mature the result is, and fit grade tells you how directly it answers the broad adult question “what did semaglutide do for weight management?”

Every STEP study with peer-reviewed or officially reported results

How to read “Difference”: a plain subtraction is labeled as such. Where the paper or FDA label supplies a model-adjusted estimated treatment difference, we label that instead. Those are not automatically the same number. A minus sign means average loss; a plus sign means average gain.

What the fit grades mean

⚠️ STEP 12 was presented at ObesityWeek 2025 and is also registered as NCT06041217. We found the official presentation and registry record, but not a full peer-reviewed primary paper as of August 4, 2026. Its row stays provisional.

Two active studies have no posted efficacy results: STEP Young (NCT05726227) and STEP TEENS Weight Maintenance (NCT06571383). Both were active and not recruiting when checked. Anyone assigning either study a weight-loss percentage before results are posted is inventing it.

The same 2.4 mg dose, nine reported averages

Here is the fastest way to see why one number cannot represent this program. The values below all came from semaglutide 2.4 mg once weekly, but from different populations, durations, designs, or analyses:

9.6% → 10.4% → 12.1% → 13.2% → 14.9% → 15.2% → 15.5% → 15.8% → 16.0%

That is a 6.4-point spread for one nominal dose. STEP 2 in adults with type 2 diabetes sits at the bottom. STEP 3 and STEP 11 tie at the top. The dose stayed the same; the study question did not.

So when someone tells you semaglutide “gets you 15%,” the honest reply is: in which trial, in which people, over how long, under which support program, and measured how?

Download the underlying data: STEP Evidence Atlas v1.0 CSV. It includes NCT number, evidence status, population, dose, result convention, fit grade, what each study answers, what it cannot answer, primary and regulatory sources, a verification note, and the last-verified date.

Why do the STEP numbers vary so much?

Answer: Five things move the number: who was enrolled, how long the study ran, what support both groups received, what question the trial was designed to answer, and how the analysis handled treatment changes and missing data. Those differences can move a headline by several percentage points without creating a contradiction.

You are not wrong to be confused

One article says 14.9%. Another says 20.7%. A third says people regained two-thirds after stopping. All three describe real findings. They just answer different questions. Most pages strip the question off and keep the number.

Here is what actually moves it.

1. Diabetes status. This is the clearest recurring pattern in the program. The current U.S. label states that Wegovy injection produced less weight reduction in patients with type 2 diabetes than in patients without it. STEP 1 versus STEP 2, same dose and same nominal length: 14.9% against 9.6%. That cross-trial contrast does not prove one biological cause for every individual.

2. Trial length. The studies in this atlas range from 44 to 104 weeks, and STEP 4 reports change from a week-20 randomization point rather than the original baseline. In STEP 5, the average curve plateaued at roughly week 60 and remained broadly stable through week 104. A 44-week result and a 104-week result do not describe the same point in treatment.

3. How much support came with the drug. STEP 3 included 30 intensive behavioral-therapy visits and an initial low-calorie phase. Most pivotal label studies used a less intensive reduced-calorie and activity program. That difference belongs next to the result, not in a footnote.

4. The study question. STEP 4 asks what happened after people who already tolerated semaglutide either continued or switched to placebo. STEP 8 asks semaglutide versus liraglutide. STEP TEENS uses change in BMI. STEP-HFpEF prioritizes symptoms and physical limitations. Those results cannot be stacked into one “best trial” leaderboard.

5. Which statistical question the analysis answered. This deserves its own section.

The two-number problem, explained without mangling the statistics

Several modern obesity trials, including STEP UP, report more than one estimand or analysis convention. The names are technical, but the distinction matters.

• The assigned-group estimate. A treatment-policy or treatment-regimen analysis starts with everyone randomized and applies prespecified rules for treatment discontinuation, dose changes, rescue treatment, and missing measurements. It is not simply the raw average of every observed participant, and the exact rules differ by study.
• The trial-product estimate. This estimates a more hypothetical question tied to continued use of the assigned trial product under the study's defined assumptions. In STEP UP it is larger than the assigned-group estimate, but that relationship is not a universal law.

STEP UP is the cleanest current example. The current U.S. label, the peer-reviewed treatment-policy analysis, and the peer-reviewed trial-product analysis give three closely related but not identical rows:

The 0.1-point differences between the current label and the paper's treatment-policy row are small reporting differences. We do not invent an explanation for them. The much larger separation between 18.7% and 20.7% comes from a different estimand, not from one source being “wrong.”

The gap reflects treatment discontinuation, dose changes, missing data, and the assumptions built into the estimand. Calling it only “the cost of side effects and quitting” is too simple.

Our rule for this page: for a current U.S. product question, we lead with the current U.S. label. When the primary paper reports another estimand, we show it separately and name it. We never blend the numbers.

Percentage points are not percent

One small distinction causes a surprising amount of confusion.

STEP 1 reported −14.9% with semaglutide and −2.4% with placebo. Plain subtraction is 12.5 percentage points. The current label's model-adjusted treatment difference is 12.4 percentage points. Neither means “12.4% better.”

📄 Keep the numbers straight

We built the downloadable STEP Evidence Atlas behind this page. Every row keeps the population, length, comparator, evidence status, fit grade, and analysis convention attached to the result.

→ Download the STEP Evidence Atlas CSV

Which STEP trial actually matches you?

Answer: The closest study is decided by age and diabetes status first, then by any specific condition, then by geography and dose — never by which trial reported the biggest number. Adults without type 2 diabetes usually start with STEP 1. Adults with type 2 diabetes start with STEP 2 or STEP UP T2D. Specific conditions like knee osteoarthritis or heart failure have their own dedicated trials.

Find yourself in this table.

Heart failure with preserved ejection fraction (HFpEF) means the heart's pumping percentage is preserved even though impaired filling and pressure can still cause heart-failure symptoms. If HFpEF is part of your question, use the condition-specific studies rather than a general obesity trial.

🔧 Which STEP Trial Is Closest to Me?

Answer six quick questions — age range, diabetes status, any specific condition, prior bariatric surgery, region, and whether you're asking about starting, continuing, or stopping. You will get:

• The one or two studies closest to your situation
• What those studies actually found, with the comparison group
• Why the match is imperfect and where it breaks down
• What that group average works out to in pounds at your starting weight
• Questions worth writing down for your clinician
• The source and the date we last checked it

This is a closest-evidence match, not a prediction of your result. No rules-based tool can tell you how your body will respond. What it can do is stop you from comparing yourself to a trial that never included anyone like you.

What did STEP 1 find, and why is 14.9% the number everyone quotes?

Answer: STEP 1 randomized 1,961 adults with overweight or obesity and no type 2 diabetes, two to one, to semaglutide 2.4 mg or placebo for 68 weeks. Average weight change was −14.9% with semaglutide and −2.4% with placebo — about 15.3 kg versus 2.6 kg. It became the reference number because it was the largest early STEP trial and studied the broad adult population most people picture when they think about weight-loss medication.

Published in the New England Journal of Medicine in February 2021. Registry number NCT03548935. Primary paper. Funded by Novo Nordisk.

Who was actually in it: average age 46, range 18 to 86. 74% women. 75% White, 13% Asian, 6% Black. 12% Hispanic or Latino. Average starting weight 105.3 kg, about 232 lb, and average BMI 37.9. People with type 2 diabetes were excluded.

How many people hit each threshold — and why you will see two sets of numbers

This is where it gets interesting, and where our denominator reconciliation adds something more useful than another copy of the headline.

The primary paper reported an available-at-week-68 analysis for participants with a body-weight measurement at that point: 1,212 in the semaglutide group and 577 in the placebo group.

The current FDA label reports a model-based treatment-regimen analysis in the randomized population, using multiple imputation for missing measurements: 1,306 assigned to semaglutide and 655 assigned to placebo.

They are both legitimate, but they are not the same analysis.

Placebo in the current label was 31.1%, 12.0%, 4.8%, and 1.7%, respectively.

The counts connect cleanly. The label says 7.2% of the semaglutide group and 11.9% of the placebo group lacked a week-68 weight measurement. Multiply the randomized counts by the corresponding observed proportions and you land on the paper's available-data counts: about 1,212 and 577. But the rate difference is not just “a bigger denominator.” The label estimates missing outcomes under its prespecified model, while the paper's quoted responder table uses participants with a week-68 measurement.

What this means for you, practically: roughly half the treated group crossed 15%, roughly a third crossed 20%, and roughly one in six did not cross 5% in the FDA label analysis. That last group is real and rarely gets the same attention as the average.

What 14.9% looks like in pounds

Illustrative arithmetic only. This shows what the reported group average or threshold equals at each starting weight. It is not a personal prediction, expected result, guarantee, or ceiling.

What STEP 1 cannot tell you

It did not test the 7.2 mg dose. It did not include adults with type 2 diabetes or anyone under 18. Its primary treatment period ended at 68 weeks. Participants who discontinued treatment were encouraged to remain in follow-up, but STEP 1 itself did not answer what happened for a full year after treatment and lifestyle support ended. The separate STEP 1 extension did.

Why did people with type 2 diabetes lose less in STEP 2?

Answer: In STEP 2, adults with type 2 diabetes on semaglutide 2.4 mg averaged −9.6% against −3.4% on placebo over 68 weeks — a 6.2-percentage-point difference. That is roughly five percentage points less weight loss than the non-diabetes population in STEP 1 on the same dose. The FDA label states directly that Wegovy produces less weight reduction in patients with type 2 diabetes.

STEP 2 also included a semaglutide 1.0 mg arm, which landed at −7.0%. Published in The Lancet in March 2021. Registry number NCT03552757. Primary paper. Recruited from 149 clinics across 12 countries. Everyone enrolled had an HbA1c between 7% and 10%.

Who was in it: average age 55, 51% women, 62% White, 26% Asian, 8% Black, 13% Hispanic or Latino. Average starting weight 99.8 kg.

Threshold results: 67.4% lost at least 5%, 44.5% lost at least 10%, 25.1% lost at least 15%, 12.8% lost at least 20% — against 30.2%, 10.2%, 4.3% and 2.3% on placebo.

A detail worth knowing if you go check this yourself

If you compare The Lancet paper to the FDA label, the numbers won't line up. The paper says 1,210 people were randomized. The label says the trial enrolled 807.

Both are right. The label only reports the 2.4 mg and placebo arms — 404 and 403 people. It leaves out the 403 people in the 1.0 mg arm, because 1.0 mg isn't a Wegovy weight-management dose. Nothing is being hidden. But if you're checking a claim and you hit that mismatch, that's why.

One thing STEP 2 does not prove

It does not establish that any individual with type 2 diabetes will lose less. It's a group average from a different population with different baseline characteristics — older, more men, more Asian participants, and all had type 2 diabetes under a protocol that permitted background glucose-lowering treatment. Comparing two separate trials tells you the averages differed. It does not tell you the reason.

Did intensive coaching change the result in STEP 3?

Answer: STEP 3 gave both groups an intensive behavioral program — 30 individual counseling visits, plus an eight-week low-calorie phase of 1,000 to 1,200 calories a day using meal replacements. The semaglutide group averaged −16.0% and the placebo group averaged −5.7%. Compared descriptively with STEP 1, the semaglutide arm differed by 1.1 points while the placebo arm differed by 3.3 points, but the two trials were not randomized against each other.

Published in JAMA in 2021. Registry number NCT03611582. Primary paper.

This is, in our view, one of the most useful and least-discussed findings in the program. Here it is with the comparison labeled honestly:

Read that bottom row again. Under the more intensive program, the placebo group lost more and the adjusted drug-versus-placebo contrast was smaller. That is a real descriptive pattern.

The honest caveat, stated plainly: STEP 1 and STEP 3 were separate trials. They were never randomized against each other. Their populations differed — STEP 3 was 81% women and 19% Black, while STEP 1 was 74% women and 6% Black. This comparison cannot prove how much of the difference came from coaching, the diet phase, the participants, chance, or another design feature.

What “real coaching” meant in the trial

The word coaching is too easy to put on a sales page. STEP 3 gives you a harder checklist:

What we would take from it: when comparing a medication-only program with one charging more for meaningful support, do not ask whether the second page says “coaching included.” Ask what the support actually consists of. STEP 3 did not prove one commercial program is better than another, but it gives you a test that marketing copy cannot pass with one vague word.

What happened when people stopped taking it?

Answer: Weight regain after withdrawal was common in both core studies that measured it. In STEP 4, people who switched to placebo after a 20-week semaglutide run-in gained 6.9% from the week-20 randomization point while those who continued lost another 7.9%. In the STEP 1 extension, an exploratory subset of 327 participants who stopped both study medication and the study lifestyle intervention regained about two-thirds of their prior semaglutide-group loss over the following year.

This is the finding that changes the decision. So we are going to be direct about it.

STEP 4: continue or switch

STEP 4 worked differently from the broad start-to-finish trials. Everyone received semaglutide for 20 weeks first. Only people who completed the run-in and reached the 2.4 mg target were eligible for randomization — 803 of the 902 who started. By week 20, the run-in population had lost an average of 10.6%.

From week 20 to week 68:

• Continued semaglutide: another −7.9%
• Switched to placebo: +6.9%
• Model-adjusted difference: 14.8 percentage points

STEP 4 primary paper · NCT03548987

The FDA label warns that this enriched design may not reflect the experience of the general population starting treatment. That warning matters. Ninety-nine of the 902 people who entered the run-in were not randomized, and 48 discontinued during the run-in because of adverse reactions. STEP 4 describes people who had already demonstrated that they could reach 2.4 mg under the protocol. It is a friendlier population than everyone walking into a telehealth intake.

The STEP 1 extension: a full year off

A subset of 327 STEP 1 participants entered an exploratory extension and were followed for a year after the trial treatment and trial lifestyle intervention ended.

STEP 1 extension primary paper

The authors summarized the group finding as regaining about two-thirds of the prior weight loss. Several cardiometabolic improvements also moved back toward baseline after withdrawal.

What this does not say: it does not say everyone regains everything. It does not say stopping is impossible or that every person ends above baseline — this subset's prior semaglutide group finished 5.6% below its original baseline. It was also an exploratory subset, not a new randomized withdrawal trial. People stop for good reasons: adverse effects, pregnancy planning, cost, access, a change in health, preference, or a clinician's decision.

🔻 Our one damaging admission

Here it is, and we would rather you hear it from us.

For weight maintenance, these medications behave more like blood-pressure treatment than a short course of antibiotics. The average effect was strongest while treatment continued, and regain after withdrawal was common. These trials do not prove that every person must stay on the same dose forever, but they do not support a three-month transformation pitch either.

Which reframes the decision. Every completed randomized comparison in this atlas favored semaglutide on the question it was designed to test. That does not guarantee your response. The harder question is whether a medically appropriate plan is sustainable for you.

Sustainability is not one thing. It is coverage or cash cost, tolerability, supply, clinical follow-up, willingness to continue, and whether the treatment remains appropriate. For insured U.S. readers, coverage is often the first financial question — not the only one.

✅ Check what your plan says before you commit

Ro's GLP-1 Insurance Coverage Checker is a free tool that currently reports coverage information for the Ozempic pen, Wegovy pen, and Zepbound pen. It can give you a personalized coverage report before you decide whether to use Ro, but the result is not a guarantee of approval, final out-of-pocket cost, or continued coverage.

→ Run Ro's free GLP-1 insurance coverage checker

If you later become a Ro Body member and a Ro-affiliated clinician prescribes treatment, Ro says its insurance team can help with benefits verification and prior-authorization paperwork when required.

Sponsored affiliate link. Ro Body is currently $39 for the first month, then $149/month on the monthly plan or as low as $74/month with an annual plan paid upfront. Medication is billed separately. Coverage, prescribing, availability, and final cost vary.

Does semaglutide still work after two years? (STEP 5)

Answer: STEP 5 followed 304 adults without type 2 diabetes for 104 weeks and found −15.2% average weight change against −2.6% on placebo, with 77.1% of the semaglutide group still down at least 5% at two years. It supports a sustained average effect over two years under trial conditions. It is a small trial and it does not appear in the Wegovy label's efficacy tables.

Published in Nature Medicine in 2022. Registry number NCT03693430. Primary paper. Participants were 77.6% women and 93.1% White, average age 47, average BMI 38.5.

Two years is real evidence and it's more than most weight-loss interventions can show. It is also not "permanent," and nobody should describe it that way. Two years is the longest dedicated randomized STEP weight-management efficacy trial. SELECT followed a different cardiovascular-risk population for longer and reported sustained weight change for up to four years, but it was not a dedicated STEP weight-management efficacy trial and should not be substituted for STEP 5's question.

How did semaglutide compare to liraglutide? (STEP 8)

Answer: STEP 8 is the only head-to-head comparison in the STEP program. Over 68 weeks, weekly semaglutide 2.4 mg produced −15.8% against −6.4% for daily liraglutide 3.0 mg — a 9.4-point difference — with pooled placebo at −1.9%. Roughly four times as many people stopped liraglutide because of adverse events as stopped semaglutide for that reason.

Published in JAMA in January 2022. Registry number NCT04074161. Primary paper. 338 participants. 94.4% finished the trial but only 80.2% finished treatment.

Head-to-head evidence is more directly informative than comparing separate trials because the treatments were evaluated inside one protocol and time period. The catch is that the semaglutide-versus-liraglutide comparison was open-label, meaning participants knew which drug they were on, which can influence behavior.

Discontinuation because of adverse events is the underrated result here. It was 12.6% with liraglutide, 3.2% with semaglutide, and 3.5% with placebo. Liraglutide is a daily injection. Semaglutide is weekly. That difference shows up in who stays on the drug.

This is not a tirzepatide comparison. No STEP trial ever tested tirzepatide (Zepbound, Mounjaro). That's a separate program — SURMOUNT — run by a different company. We compare them on a separate page.

What did STEP 6, STEP 7, STEP 11, and STEP 12 add?

Answer: These four studies extended the evidence into East and Southeast Asian populations using local eligibility thresholds and different mixes of type 2 diabetes, geography, and follow-up. Main semaglutide 2.4 mg results ranged from −12.1% to −16.0%. Their value is direct study in populations that were less represented in STEP 1 — not permission to predict an individual's result from ancestry.

STEP 6 paper · STEP 7 paper · STEP 11 paper · STEP 12 official presentation

STEP 11's responder distribution

In STEP 11, 96% of the semaglutide group lost at least 5%, 78% lost at least 10%, and 53% lost at least 15%, versus 25%, 10%, and 4.2% with placebo.

Among the directly reported ≥5% body-weight responder rates we verified for this atlas, STEP 11's 96% is the highest. It is not the strongest result at every higher threshold: STEP UP's current label reports higher proportions reaching at least 10% and at least 15% with 7.2 mg. This is why “strongest in the program” is too vague to be useful.

STEP 11 used BMI ≥25 as the trial's local obesity criterion. That threshold is part of the study definition; it should not be silently replaced with the higher BMI floors common in U.S. trial summaries.

A population detail that changes how STEP 6 reads

STEP 6 was 63% men, with an average starting weight of 87.5 kg and average BMI of 31.9. STEP 1 was predominantly women and began at a higher average body weight and BMI. Those are materially different study populations before anyone reaches for an ethnicity-based explanation.

⚠️ STEP 12 remains provisional. Its official presentation reports a treatment-policy mean of −12.1% versus −2.2% and a trial-product mean of −12.7% versus −2.4%. We did not find a full peer-reviewed primary paper as of August 4, 2026.

Do not turn these studies into an ethnicity calculator. Population evidence tells you where the drug was directly studied. It does not predict what one person will do.

What did the specialized STEP trials find?

Answer: STEP 9, STEP 10, STEP-HFpEF, STEP-HFpEF DM, and BARI-STEP studied semaglutide in people with a specific condition or treatment history alongside obesity. They add evidence that STEP 1 cannot supply, but their results belong with the condition-specific outcome and population — not in a general “which trial lost the most?” ranking.

STEP 9 — obesity and knee osteoarthritis

STEP 9 randomized 407 adults with obesity and moderate knee osteoarthritis with at least moderate pain. The trial ran 68 weeks across 61 sites in 11 countries. Average age was 56 and average BMI was 40.3.

• Weight: −13.7% with semaglutide versus −3.2% with placebo
• WOMAC pain score: −41.7 points versus −27.5 points
• SF-36 physical-function score: +12.0 points versus +6.5 points

Primary paper · NCT05064735

What it cannot separate: whether pain improvement came from weight loss, another drug-mediated pathway, changes in activity, or a combination. The trial showed a treatment effect on both outcomes; it did not isolate the mechanism.

STEP 10 — obesity and prediabetes

STEP 10 randomized 207 adults for 52 weeks of treatment followed by a 28-week off-treatment period.

• Weight: −13.9% with semaglutide versus −2.7% with placebo
• Normoglycemia at week 52: 81.1% versus 14.1%

Primary paper · NCT05040971

That 81.1% is striking and it is an on-treatment week-52 result under the trial's definition. Returning to normoglycemia at that point is not the same claim as permanently preventing type 2 diabetes. For the practical access and eligibility questions, see our separate GLP-1 for prediabetes guide.

STEP-HFpEF and STEP-HFpEF DM — obesity-related heart failure

These two 52-week trials were split by type 2 diabetes status. The co-primary outcomes focused on heart-failure symptoms and physical limitations measured with the Kansas City Cardiomyopathy Questionnaire clinical summary score, plus body weight — not weight alone.

STEP-HFpEF paper · STEP-HFpEF DM paper

The lower average weight change in the diabetes trial repeats a pattern seen elsewhere, but comparing these two separate populations does not prove one causal explanation.

If you have HFpEF, this is a cardiology conversation, not a comparison-shopping question. These trials were not powered to settle every hard clinical-event question, and this page does not turn symptom-score evidence into a personal treatment recommendation.

BARI-STEP — after bariatric surgery

BARI-STEP randomized 70 people who had a suboptimal clinical response or weight regain at least one year after gastric bypass or sleeve gastrectomy. The intention-to-treat set included 63 people.

The model-adjusted treatment difference was −19.1 percentage points. That is larger than the simple observed subtraction of 18.4 points, which is why this page keeps “observed” and “adjusted” attached to the number.

Primary paper · NCT05073835

BARI-STEP is also the funding exception that makes “Novo Nordisk designed and ran every STEP study” false. It was investigator led and sponsored by University College London; Novo Nordisk supplied funding and trial product alongside public and charitable support, while the paper states that the company was not involved in trial design, data analysis, interpretation, or manuscript preparation.

The result is important. It is also the smallest randomized study in the atlas and applies to a very specific post-surgical population. It is a weak basis for a general semaglutide expectation and a strong reason to involve a bariatric team when surgery history is part of the question.

What did STEP TEENS show, and what is still being studied in children?

Answer: STEP TEENS randomized 201 adolescents aged 12 to under 18 and reported a 16.1% reduction in BMI with semaglutide versus a 0.6% increase with placebo at 68 weeks. Because the primary endpoint was percentage change in BMI rather than adult percentage body-weight change, it cannot be placed in a “which trial lost the most?” ranking.

Published in the New England Journal of Medicine in 2022. Registry number NCT04102189. Primary paper. Average age was 15.4, 38% were male, and average starting weight was 107.5 kg.

Why BMI rather than only weight percentage: adolescents are still growing, so pediatric obesity studies use BMI-based outcomes to account for height as well as weight. That does not make the result “better” or “worse” than an adult percentage; it makes it a different endpoint.

One safety difference worth knowing: cholelithiasis, or gallstones, was reported in 3.8% of semaglutide-treated adolescents versus 0% with placebo in the current U.S. label. Cholecystitis was reported in 0.8% versus 0%. Those events are one reason a pediatric result should never be reduced to the BMI headline alone.

Still running, no efficacy results posted when checked:

• STEP Young — active, not recruiting; includes younger children and longer-term safety follow-up
• STEP TEENS Weight Maintenance — active, not recruiting; a long-term adolescent maintenance study

If a page assigns either ongoing study an efficacy percentage before results are posted, the number is not from that trial.

If you are researching for a child, the only next step from this section is a qualified pediatric clinician. We are not going to place a provider or affiliate button under pediatric efficacy data, and you should be skeptical of a page that does.

What did STEP UP show about the 7.2 mg dose?

Answer: STEP UP randomized 1,407 adults with obesity and no diabetes to semaglutide 7.2 mg, semaglutide 2.4 mg, or placebo for 72 weeks. The current U.S. label reports −18.8%, −15.5%, and −3.9%, respectively. FDA approved the 7.2 mg dose on March 19, 2026 for certain adults. The higher dose produced more average weight loss and a much higher rate of altered skin sensation.

STEP UP primary paper · NCT05646706 · FDA approval announcement

STEP UP T2D used the same three-arm structure in 512 adults with obesity and type 2 diabetes. Using the current U.S. label values:

• 7.2 mg: −13.2%
• 2.4 mg: −10.4%
• Placebo: −3.8%

The peer-reviewed treatment-policy report displays placebo as −3.9% and an adjusted 7.2 mg-versus-placebo difference of 9.3 points; the current U.S. label displays −3.8% and 9.4 points. We use the current label in the master table and keep the paper-versus-label difference explicit rather than silently swapping them.

STEP UP T2D primary paper · NCT05649137

Why the label says 18.8% while the paper says 18.7% or 20.7%

The current U.S. label and the paper do not display the result in exactly the same way:

We lead with 18.8% because this is a U.S. page describing the currently approved product. We preserve 20.7% because it is a legitimate trial-product estimate, not because it is the better headline.

In the published treatment-policy analysis, 31.2% of the 7.2 mg group reached at least 25% weight loss. In the trial-product analysis, 33.2% did. “About a third” is fair; “everyone can lose a quarter of their body weight” is not.

Who was in STEP UP: average age 47, range 18 to 80. 74% women, 86% White, 9% Black, 4% Asian, and 5% Hispanic or Latino. Average starting weight was about 113 kg and average BMI 39.9.

The under-covered 7.2 mg safety signal: dysesthesia

The current U.S. label uses dysesthesia as an umbrella for altered skin sensations such as burning, tingling, prickling, pain from light touch, or skin sensitivity.

It was reported by:

• 22% of the 7.2 mg group
• 6% of the 2.4 mg group
• 0.3% of the placebo group

Among 288 people who experienced dysesthesia with 7.2 mg:

• 23% had a dose reduction
• 8% had treatment temporarily interrupted
• 2% discontinued permanently
• 18% had not reported recovery by the end of the trial
• 38 people later increased back to 7.2 mg after resolution; 17, or 45%, had recurrence

The label says dose reduction or interruption was associated with a shorter time to resolution. That is useful clinical-context information. It is not permission to change a dose without the prescriber.

Hair loss also differed sharply by sex in the label.

This section is not a dosing guide. The current label permits escalation above 2.4 mg only after at least four weeks at 2.4 mg and when additional weight reduction is clinically indicated. Only a prescriber who knows the patient's history can decide whether that applies.

The trap in the FDA label: the study numbers are a separate system

Answer: The current Wegovy prescribing information does not use the public STEP names in its clinical-study tables. It uses an internal sequence that includes SELECT, OASIS 4, STEP TEENS, and ESSENCE while omitting several numbered STEP trials. STEP 1 happens to be Study 2 and STEP 2 happens to be Study 3, but the sequence is not a uniform one-number shift.

Here is the crosswalk from the U.S. prescribing information revised 06/2026:

Current U.S. Wegovy prescribing information

Why this is not a trivia point

The label's safety sections cite these internal study numbers. Read them as public STEP numbers and you can attach a warning to the wrong population.

• The hypoglycemia warning cites Study 3. In the label, Study 3 is STEP 2, the type 2 diabetes trial.
• The diabetic-retinopathy warning also cites Study 3 for the same reason.
• The pooled hair-loss figures for Studies 2, 3, and 4 refer to STEP 1, STEP 2, and STEP 3.
• The older anti-drug antibody discussion for Studies 2 and 3 refers to STEP 1 and STEP 2.
• Label Study 7 is OASIS 4, an oral-tablet study. It is not STEP 7, the 44-week East Asian injection trial.

A reader who assumes “Study 3” means STEP 3 can conclude that a diabetes-specific warning came from the intensive-behavioral-therapy trial in people without diabetes. It did not.

Seven numbered STEP trials are absent from the adult efficacy tables

The adult weight-management efficacy sections of the current label include STEP 1, STEP 2, STEP 3, STEP 4, STEP 6, STEP UP, and STEP UP T2D. STEP TEENS appears in the pediatric efficacy section.

Numbered STEP 5, STEP 7, STEP 8, STEP 9, STEP 10, STEP 11, and STEP 12 do not appear in those adult efficacy tables.

That does not make the omitted studies unreal, weak, or unreviewed. It means the FDA label is a regulatory document, not a complete bibliography of every later trial. If someone says STEP 5's two-year result or STEP 8's liraglutide comparison is “in the Wegovy label,” the current label does not support that claim.

One more label-reading trap

The label reports anti-semaglutide antibodies in 3% of participants in the older injection studies and 15.4% in the newer 7.2 mg studies. That looks like a five-fold jump.

It is not a valid cross-study comparison. The label says the newer studies used a different assay and that assay differences preclude comparing those incidence figures. Different test, different number. This is exactly how a scary headline gets manufactured from two values the source tells you not to compare.

What did participants get besides the drug?

Answer: Semaglutide was tested as an adjunct to diet and activity intervention, not as a pill-or-pen dropped into an otherwise empty protocol. The exact support package varied, and the current FDA label does not establish that every one of the 18 studies used an identical program. It does show that the pivotal label studies paired treatment with structured calorie, activity, and counseling requirements.

Directly from the current label:

Current U.S. Wegovy prescribing information

Across the wider program, the intervention details differ by population and trial. The honest rule is not “every participant got exactly the same package.” It is: the headline percentages came from protocols that included more than medication.

The support-fidelity test

When a commercial program says its expected results are “based on STEP,” ask six questions:

1. How many live human visits are included?
2. Who delivers the nutrition or behavior support, and what are their credentials?
3. Is there a written calorie or food plan, or only generic articles?
4. Is there a specific activity target and progression plan?
5. Does anyone review adherence, adverse effects, and a plateau?
6. Does support continue after the first month, or does it turn into refill-only messaging?

A telehealth program may provide excellent support, light support, or almost none. Do not assume either direction. Verify the actual service. When you compare your progress with 14.9%, remember that the trial result includes the trial's support structure as well as the medication.

Who stopped treatment, and why?

Answer: Discontinuation is part of the efficacy story, not a footnote. In the pooled 2.4 mg adult safety trials used by the current label, 16% of semaglutide participants and 19% of placebo participants permanently discontinued study treatment; 6.8% versus 3.2% stopped because of adverse reactions. Gastrointestinal reactions were the most common pattern.

The discontinuation ledger

The rows are not fully interchangeable. STEP 4's 5.3% occurred during an open-label run-in before randomization. STEP 8 compared active products in an open-label treatment comparison. The pooled label row combines three placebo-controlled trials.

Common adverse reactions in the adult 2.4 mg weight-management trials

From the current U.S. label:

Most gastrointestinal events were mild to moderate and occurred most often during dose escalation. “Common” does not mean “trivial” for every person: nearly three in four reported a gastrointestinal reaction and about one in 25 had a severe gastrointestinal reaction under the label's definition.

Wegovy carries a boxed warning about thyroid C-cell tumors observed in rodents. It is contraindicated in patients with a personal or family history of medullary thyroid carcinoma or in patients with Multiple Endocrine Neoplasia syndrome type 2. Those are hard contraindications in the current label, not marketing copy. Read the current prescribing information for the complete boxed warning, contraindications, precautions, and medication guide.

What a trial program cannot settle well: very rare risks. Randomized trials are designed to measure common and prespecified outcomes, not to guarantee detection of every event that might emerge after much wider use. That is why postmarketing surveillance and current label revisions matter.

Do the STEP results apply to compounded semaglutide?

Answer: No. The STEP studies evaluated specific semaglutide products, doses, and protocols. Those findings do not establish the safety, effectiveness, quality, bioavailability, or equivalence of a compounded finished product. FDA states that compounded drugs are not FDA-approved and that FDA does not verify their safety, effectiveness, or quality before marketing.

This is where a lot of pages get sloppy, so we are going to be exact.

Compounded medication is prepared under federal and state compounding frameworks, commonly by a state-licensed pharmacy or an FDA-registered outsourcing facility, to meet a clinical need that the prescriber identifies. It is a different regulatory category from an FDA-approved finished drug.

What the STEP studies tested was a defined investigational or approved semaglutide product, made to specified manufacturing standards, given at specified doses and schedules, to participants screened and monitored under a protocol.

What that evidence covers is that studied product and protocol. It does not automatically transfer to a different finished product because the ingredient name on a page looks familiar.

So if an ad places 14.9%, 18.8%, or 20.7% next to a compounded offer, the percentage may be a real STEP result, but it was not generated by a trial of that compounded finished product.

We are not making a blanket safety judgment about every compounded prescription. We are drawing the evidence boundary that the source requires. FDA-approved product evidence and compounded-product evidence are not interchangeable.

FDA has also warned about dosing errors and fraudulent products marketed as compounded GLP-1 drugs. Those warnings do not prove that every legitimate compounded prescription has the same problem; they are reasons to verify the prescriber, dispensing source, concentration, units, instructions, and legal basis rather than treating “compounded” as one uniform product.

Compounding and the FDA: Questions and Answers · FDA's concerns with unapproved GLP-1 drugs

There is no provider CTA in this section. A section explaining that STEP evidence does not transfer to compounded finished products is not the place to sell one.

What about Ozempic, the Wegovy pill, or tirzepatide?

Answer: No STEP trial evaluated Ozempic as the marketed diabetes product, although STEP 2 did include a semaglutide 1.0 mg arm. Oral Wegovy was developed in the OASIS program. Tirzepatide is a different molecule — a dual GIP/GLP-1 receptor agonist — developed for weight management in the SURMOUNT program. Same broad treatment landscape, different products and evidence.

Ozempic's current maximum recommended dose is 2.0 mg once weekly and its FDA indications are tied to type 2 diabetes and related cardiovascular or kidney-risk uses, not the Wegovy weight-management label. Using “semaglutide was in STEP” to erase the product, formulation, dose, and indication is exactly the shortcut this page is built to stop.

The program dictionary

One naming collision to watch: a registry entry called STEP-HFpEF DM ORAL (NCT07390110) is an observational target-trial emulation using existing records. It is not another completed randomized STEP efficacy trial, and no results were posted when checked. A familiar name does not make the design the same.

For medication-to-medication decision making, compare the correct trial programs side by side rather than borrowing STEP 1's number for every semaglutide product or using SURMOUNT as though it were a STEP trial.

Are STEP trial results the same as real-world results?

Answer: No — but there is no single predictable “real-world discount.” Persistent, well-supported cohorts that reach treatment doses can approach STEP-like averages, while all-starter populations are pulled down by discontinuation, lower doses, access gaps, and missing follow-up. Before comparing any real-world number with STEP 1, ask who stayed in the denominator.

The denominator-first comparison

Academic-clinic study · MassHealth study · SHAPE study

This table explains why two real-world articles can appear to disagree while both are accurate. One starts with everyone who initiated and asks whether they stayed. Another restricts the analysis to people who remained persistent for a year and asks how much those remaining patients lost.

The four questions to ask before believing a real-world percentage

1. Who entered the denominator — every starter or only persistent users?
2. Was the result semaglutide-only, or were multiple drugs and switchers combined?
3. How many people reached the studied maintenance dose?
4. Whose weight was still available at the reported time point?

Randomized trials and observational data are both useful. STEP tells you what happened under a defined protocol and analysis plan. Real-world studies tell you what happened inside a specific health system, coverage arrangement, persistence definition, and missing-data pattern. The source of the difference is not always “willpower,” and it is not always cost alone.

What can the STEP trials still not tell you?

Answer: The STEP program provides strong randomized evidence, but it has real boundaries: most pivotal studies were manufacturer sponsored, eligibility criteria narrowed who participated, representation was uneven, rare harms remain difficult to measure, and the longest dedicated randomized STEP weight-management trial ran for two years. Those limits do not erase the findings. They define what you can honestly conclude.

Funding. Most pivotal STEP trials were designed and funded by Novo Nordisk, and company employees were involved in many trial reports. That is standard for registrational drug development and remains a material fact. BARI-STEP is the exception in this atlas: it was investigator led and UCL sponsored, with Novo Nordisk funding and trial-product support but no reported role in design, analysis, interpretation, or manuscript preparation.

Age. Across the current label's adult weight-management trials, 233 participants, or 9%, were age 65 to under 75 and 23, or 1%, were 75 or older. If you are 75 or older, the dedicated adult weight-management trial evidence represented in that pooled label group is 23 people.

Race and ethnicity. The pooled 2.4 mg adult safety population in the current label was 72% White, 14% Asian, 9% Black or African American, and 5% other or unknown. Individual trials differed sharply. The point is not to turn a demographic row into an outcome calculator; it is to see where direct representation was thin.

Duration. STEP 5 is the longest dedicated randomized STEP weight-management efficacy study at 104 weeks. SELECT reported sustained weight change for up to four years in a different cardiovascular-outcomes population and question. “Two years is all the semaglutide weight data” is false; “two years is the ceiling of dedicated randomized STEP weight-management efficacy” is accurate.

Rare risks. Trial programs are not large enough to reliably characterize every very rare event. Current labeling and postmarketing surveillance remain part of the evidence after approval.

Cross-trial comparisons are descriptive, not randomized. Every time this page compares STEP 1 with STEP 2, STEP 3, STEP 11, or another separate study, the comparison is an observation across protocols and populations. It does not isolate one causal reason for the difference.

Averages hide people. A 14.9% mean contains people above 30%, people near the mean, and people below 5%. The responder distribution belongs next to the average.

The trials do not tell you which commercial provider is best. A clinical result cannot verify a telehealth service's price, clinician access, support quality, cancellation process, dispensing source, or whether that service can reproduce the study protocol.

How to use all of this with your doctor

Answer: Use the closest trial to frame a better conversation, not to set a guaranteed target. Five questions turn this page into something useful in a short appointment.

1. Which trial population is closest to me? Bring the answer. It changes which group estimate is relevant; it does not create a personal expected range.
2. Is the number I've seen the "everyone" figure or the "if everyone stuck with it" figure?
3. What share of people hit each threshold? Not just the average — the spread.
4. How long was treatment actually studied at this dose?
5. What's the plan if my coverage changes or I need to stop?

Questions worth writing down

• Which trial best matches my medical situation, and what did it find?
• Which dose are we starting at, and what's the plan for increasing it?
• What result would count as working, for me specifically?
• What side effects should make me call you rather than wait?
• How will we decide whether to continue at six months?
• What happens if my insurance stops covering it?
• What would make stopping the right call medically?

A one-minute micro-commitment before the appointment

Choose one trial, write down its population and main result, then circle one reason you are not a perfect match. That single step is more useful than bringing a screenshot that says “15%.”

• Closest study: __________
• Why it is close: __________
• Why it is imperfect: __________
• The question I need answered: __________

If coverage or prior authorization is the sticking point, use our separate GLP-1 prior-authorization guide rather than trying to solve an insurance process inside a clinical-trial table.

Does this evidence fit your situation well enough to act on? Get a personalized GLP-1 action plan based on whether you are exploring FDA-approved treatment, insurance coverage, cash-pay care, or questions for a clinician you already have.

How can you get the FDA-approved medication studied in STEP?

Answer: The STEP injection program studied Novo Nordisk semaglutide products that support the FDA-approved Wegovy evidence base. Wegovy is now available in the United States as injection, including the 7.2 mg dose for certain adults, and as an oral tablet under its own evidence and labeling. Your cost depends on the prescribed product and dose, insurance, manufacturer programs, pharmacy channel, and any separate telehealth membership fee.

If you have read this far, you already know the harder decision is not whether a group average exists. It is whether you can obtain medically appropriate treatment and sustain it.

The honest trade-off with Ro

Ro charges a care membership in addition to the medication. If you already have a clinician who will prescribe, monitor treatment, and handle prior authorization, a direct manufacturer or pharmacy route may involve lower care fees. Ro's value is bundling online clinical evaluation, ongoing provider messaging, insurance navigation, prior-authorization support when required, tracking tools, and coaching into one service.

That is a useful bundle for someone who does not already have those pieces. It is unnecessary duplication for someone who does.

What Ro currently publishes

Verified August 4, 2026. Ro charges the care plan and medication separately.

The promotional dates, dose prices, multi-month savings, shipping terms, and eligibility rules can change. Confirm the prescribed dose, medication charge, care-plan charge, and renewal terms on Ro's current pricing page before paying.

Ro's free checker currently covers the Ozempic pen, Wegovy pen, and Zepbound pen. Ro states that it cannot currently check insurance coverage for the Wegovy pill, Foundayo pill, or Zepbound KwikPen through that free tool. The checker provides a personalized report; it does not guarantee approval, a particular copay, or continued coverage.

Ro says its insurance team can check benefits and help with prior-authorization paperwork after a Ro-affiliated clinician determines that treatment is appropriate. Assistance does not guarantee approval or a specific copay.

✅ See what the FDA-approved route would actually cost you

Start with coverage if you have insurance. Start with dose-specific cash pricing if you do not. Either way, compare the medication charge and care membership separately.

→ Check Wegovy coverage with Ro's free tool
→ See Ro's current weight-loss pricing

Sponsored affiliate links. Clinical eligibility, prescribing, medication availability, insurance approval, and final cost vary. Medication is not included in the Ro Body membership fee.

How we built this STEP trials summary

Answer: We extracted study design and outcomes from primary publications, ClinicalTrials.gov records, current U.S. prescribing information, and official congress material, then normalized population, dose, duration, comparator, endpoint, analysis convention, evidence status, and practical fit. We did not reanalyze patient-level data, and we did not fill missing evidence with a confident-sounding guess.

Who we are. Weight Loss Provider Guide is an independent comparison resource for GLP-1 telehealth providers. This page was produced by Weight Loss Provider Guide. No physician review is represented on this page.

Why this page exists. Trial headlines routinely detach a percentage from the people, time point, comparator, endpoint, and estimand that produced it. This evidence atlas keeps those conditions attached so a group average cannot quietly turn into a personal promise.

Our source order

1. Current FDA labeling and FDA approval materials
2. Peer-reviewed primary trial publications
3. ClinicalTrials.gov registration and posted results
4. Official congress presentations, clearly labeled provisional
5. Manufacturer material only when it is the first-party source for a commercial fact or no stronger trial source is yet available

What we did ourselves

• Built the 18-row STEP Evidence Atlas using the same fields for every study
• Added an evidence-status ledger so peer-reviewed, label-reported, congress-reported, and ongoing studies are not presented as equal
• Added fit grades to stop broad adult, withdrawal, pediatric, head-to-head, and specialized studies from being ranked together
• Built the FDA-label-to-public-trial crosswalk by matching NCT numbers rather than guessing from study numbers
• Reconciled STEP 1's available-at-week-68 responder table with the current label's randomized-population analysis and missing-weight counts
• Separated simple arithmetic differences from model-adjusted estimated treatment differences
• Compared STEP 1 and STEP 3 descriptively while labeling that comparison as cross-trial rather than causal
• Built the denominator-first real-world table so persistence-selected outcomes cannot be mistaken for all-starter outcomes
• Created a downloadable, spreadsheet-ready data file with the source and the question each study can and cannot answer

Fit-grade methodology

The grades describe comparability to a broad adult weight-management question, not scientific quality or personal suitability.

Calculation rules

• Simple arithmetic gap: reported semaglutide group mean minus reported comparison-group mean.
• Adjusted difference: the paper's or label's model-adjusted estimated treatment difference.
• Pounds illustrations: starting weight multiplied by the reported percentage.
• Responder rates: copied from the stated analysis and denominator; available-case and imputed randomized-population results remain separate.
• No personal forecast: no trial mean is converted into “what you should expect.”

What we did not do

We did not reanalyze patient-level data. We did not use patient testimonials to support efficacy or safety. We did not transfer FDA-approved-product evidence to compounded finished products. We did not mark STEP 12 peer-reviewed when we could not verify a full primary paper. We did not turn ongoing pediatric studies into estimates.

Download the data: STEP Evidence Atlas v1.0 CSV

Version history

• v1.0 — August 4, 2026: first published version
• Material corrections and evidence-status changes will be logged here rather than silently overwritten.

STEP trials FAQ

Answer: These are the short answers to the questions most likely to send someone back to search. Each one keeps the condition that makes the answer accurate.

What does STEP stand for?

Semaglutide Treatment Effect in People with obesity. The name began with Novo Nordisk's phase 3 weight-management program and now appears across numbered, pediatric, dose-expansion, heart-failure, and investigator-led studies.

How many STEP trials are there?

There are 12 numbered trials in this atlas plus named studies. Eighteen studies have results represented here: 17 have a peer-reviewed primary report or current FDA-label result, and STEP 12 has an official congress report but no full peer-reviewed primary paper that we could verify as of August 4, 2026. STEP Young and STEP TEENS Weight Maintenance remain active without posted efficacy results.

Which STEP trial produced the 14.9% figure?

STEP 1, in 1,961 adults with overweight or obesity and no type 2 diabetes over 68 weeks on semaglutide 2.4 mg weekly. Placebo in the same trial was −2.4%.

Is 15% an average or a maximum?

An average, not a cap. In STEP 1's current FDA-label analysis, 47.9% reached at least 15%, 30.2% reached at least 20%, and 16.5% did not reach 5%.

Why did people with type 2 diabetes lose less in STEP 2?

STEP 2 averaged −9.6% versus STEP 1's −14.9% at the same 2.4 mg dose and nominal 68-week length. The current label states that Wegovy injection produced less weight reduction in patients with type 2 diabetes. Comparing separate trials shows a recurring group pattern; it does not establish one biological reason for every individual.

Why do I see 12.4 and 12.5 points for STEP 1?

Because they are different calculations. Subtracting the displayed group means, 14.9 minus 2.4, gives 12.5 percentage points. The FDA label's model-adjusted estimated treatment difference is 12.4 points.

Why do I see 18.8%, 18.7%, and 20.7% for STEP UP?

The current U.S. label reports 18.8%. The peer-reviewed treatment-policy analysis reports 18.7%. The peer-reviewed trial-product analysis reports 20.7%. The first two are a small label-versus-paper reporting difference; 20.7% answers a different, more hypothetical estimand.

What happened when participants stopped semaglutide?

In the STEP 1 extension subset, the prior semaglutide group regained 11.6 percentage points over the following year and remained 5.6% below original baseline. In STEP 4, people switched to placebo gained 6.9% over 48 weeks while those who continued semaglutide lost another 7.9% from the week-20 randomization point.

Did everyone regain two-thirds of the weight?

No. “Two-thirds” summarizes the average regain in an exploratory subset of 327 people. It was not a rule followed by every participant.

Was Ozempic tested in STEP?

Not as the marketed Ozempic product. STEP 2 included a semaglutide 1.0 mg arm, but the STEP program evaluated semaglutide under weight-management protocols and the Wegovy evidence path. Product, dose, formulation, and indication still matter.

Do STEP results apply to compounded semaglutide?

No. The trials do not establish the safety, effectiveness, quality, bioavailability, or equivalence of a compounded finished product. Compounded drugs are not FDA-approved.

Are all STEP trials independent of Novo Nordisk?

No. Most pivotal STEP trials were Novo Nordisk sponsored. BARI-STEP was investigator led and UCL sponsored, with Novo Nordisk funding and trial-product support but no reported role in design, analysis, interpretation, or manuscript preparation.

Which STEP studies appear in the current FDA label?

The adult efficacy sections include STEP 1, STEP 2, STEP 3, STEP 4, STEP 6, STEP UP, and STEP UP T2D under the label's internal study numbers. STEP TEENS appears in the pediatric section. Numbered STEP 5, STEP 7, STEP 8, STEP 9, STEP 10, STEP 11, and STEP 12 do not appear in the adult efficacy tables.

What is the newest STEP evidence?

The FDA approved Wegovy 7.2 mg on March 19, 2026 based on STEP UP and STEP UP T2D. BARI-STEP has a 2026 peer-reviewed paper. STEP 11 was peer reviewed in 2025, while STEP 12 remains officially reported but provisional in this atlas. The pediatric ongoing-study statuses were last checked August 4, 2026.

Which STEP trial is closest to my situation?

Use the closest-trial matcher. It returns the nearest study, why it matches, where it breaks down, and the questions worth taking to a clinician. It is a closest-evidence match, not a forecast.

Still not sure which GLP-1 program is right for you?

Take our free 60-second matching quiz.

You will get a personalized action plan based on where you actually are — whether you are exploring FDA-approved treatment, trying to get insurance coverage, paying cash, or preparing questions for a clinician you already have.

Sources

Current regulatory sources

• Wegovy U.S. Prescribing Information, revised 06/2026
• Ozempic U.S. Prescribing Information, revised 05/2026
• FDA approval of Wegovy 7.2 mg, March 19, 2026
• Compounding and the FDA: Questions and Answers
• FDA's concerns with unapproved GLP-1 drugs used for weight loss

Primary trial publications and records

• SELECT long-term weight analysis — Ryan et al., Nature Medicine, 2024 · NCT03574597
• STEP 1 — Wilding et al., New England Journal of Medicine, 2021 · NCT03548935
• STEP 1 extension — Wilding et al., Diabetes, Obesity and Metabolism, 2022
• STEP 2 — Davies et al., The Lancet, 2021 · NCT03552757
• STEP 3 — Wadden et al., JAMA, 2021 · NCT03611582
• STEP 4 — Rubino et al., JAMA, 2021 · NCT03548987
• STEP 5 — Garvey et al., Nature Medicine, 2022 · NCT03693430
• STEP 6 — Kadowaki et al., The Lancet Diabetes & Endocrinology, 2022 · NCT03811574
• STEP 7 — Mu et al., The Lancet Diabetes & Endocrinology, 2024 · NCT04251156
• STEP 8 — Rubino et al., JAMA, 2022 · NCT04074161
• STEP 9 — Bliddal et al., New England Journal of Medicine, 2024 · NCT05064735
• STEP 10 — McGowan et al., The Lancet Diabetes & Endocrinology, 2024 · NCT05040971
• STEP 11 — Lim et al., The Lancet Diabetes & Endocrinology, 2025 · NCT04998136
• STEP 12 official ObesityWeek 2025 presentation · NCT06041217
• STEP TEENS — Weghuber et al., New England Journal of Medicine, 2022 · NCT04102189
• STEP UP — Wharton et al., The Lancet Diabetes & Endocrinology, 2025 · NCT05646706
• STEP UP T2D — Lingvay et al., The Lancet Diabetes & Endocrinology, 2025 · NCT05649137
• STEP-HFpEF — Kosiborod et al., New England Journal of Medicine, 2023 · NCT04788511
• STEP-HFpEF DM — Kosiborod et al., New England Journal of Medicine, 2024 · NCT04916470
• BARI-STEP — Stanley et al., Nature Medicine, 2026 · NCT05073835

Ongoing studies

• STEP Young — NCT05726227
• STEP TEENS Weight Maintenance — NCT06571383
• STEP-HFpEF DM ORAL — NCT07390110

Real-world evidence used in the denominator-first table

• Samuels et al., academic obesity clinic, 2025
• Meyer et al., MassHealth, 2026
• Ng et al., SHAPE, 2025

Commercial facts

• Ro Body membership and weight-loss pricing
• Ro GLP-1 Insurance Coverage Checker

Weight Loss Provider Guide is an independent comparison resource for GLP-1 telehealth providers. We may earn compensation from some providers featured elsewhere on this site. No provider paid to be included in, excluded from, or favored in this clinical-trial summary, and no compensation influenced any figure, fit grade, evidence-status label, or trial assessment on this page.

This page summarizes published and officially reported clinical evidence. It is not medical advice and does not replace current FDA-approved prescribing information or a conversation with a licensed clinician. Only a prescriber who knows your medical history can determine whether a medication or dose is appropriate.