STEP TEENS is quoted at adults constantly, usually as “16%”, and the figure is real. It is also not the same kind of number as the percentages from adult trials like STEP 1, because the trial’s preregistered primary endpoint is a change in body-mass index, not a change in body weight — and in a fifteen-year-old the height in that ratio’s denominator does not hold still. What is licensed in this age group, how the adolescent dose was arrived at, and what the longer liraglutide evidence shows are all covered in the adolescents article. This page is about how the trial itself was built and what its own tables say.
What was run
From October 2019 through July 2020, 229 adolescents were screened, and everyone who passed screening entered a 12-week lifestyle intervention run-in before any randomization happened, following regulatory guidance for pediatric obesity trials. Participants who completed the run-in but no longer met the body-mass-index criteria were recorded as ineligible at screening.201 were randomized, 134 to semaglutide and 67 to placebo.[1]
Eligibility ran from 12 to under 18 years of age, with a body-mass index at or above the 95th percentile for sex and age, or at or above the 85th percentile with at least one weight-related coexisting condition, and at least one unsuccessful dietary attempt at weight loss behind them. Randomization was 2:1 through an interactive web-response system, stratified by sex and by pubertal status — Tanner stage 2 or 3 against stage 4 or 5. The semaglutide dose was escalated over 16 weeks from 0.25 mg to 2.4 mg, or to the highest dose that did not cause unacceptable adverse effects. Treatment ran 68 weeks; a 7-week follow-up period with no injections followed it, to week 75. Both groups received behavioral lifestyle therapy throughout, with parents or guardians included.[1]
The preregistered primary endpoint was the percentage change in body-mass index from baseline to week 68. The secondary confirmatory endpoint was weight loss of at least 5% at week 68. The registry entry for NCT04102189 lists the percentage change in body-mass index as the sole primary outcome.[2] Statistical testing used a hierarchy, and the trial notes that supportive and exploratory endpoints were not controlled for multiplicity and should not be used to infer definitive treatment effects.[1]
Who enrolled: mean age 15.4, mean body weight 107.5 kg, mean body-mass index 37.0, 62% girls, 79% recorded as White, with hypertension in 13% and type 2 diabetes in 4%. All but one participant had obesity rather than overweight.[1]
The result, and the three versions of it
Under the treatment-policy estimand — the primary one, applied to everyone randomized regardless of adherence or rescue treatment — body-mass index changed −16.1% on semaglutide against +0.6% on placebo, an estimated difference of −16.7 percentage points (95% CI, −20.3 to −13.2), at P < 0.001.[1] The registry posts the observed means instead: −16.2% (SD 12.9) against −0.1% (SD 8.6), with a difference of −16.75 (95% CI, −20.27 to −13.23).[2] A trial-product estimand, which assumes everyone took the assigned regimen without rescue, returns a third set again.
All three describe the same trial and none of them is the others. The one thing they agree on is the size of the gap, roughly sixteen and a half percentage points of body-mass index, with an interval running from about twenty down to about thirteen.
The two rulers do not measure the same thing
Here is the number that explains why an adolescent percentage cannot be laid next to an adult one. In the same trial, over the same 68 weeks, the registry’s observed mean change in body weight was −14.8% (SD 13.2) on semaglutide and +2.3% (SD 9.1) on placebo — a gain of 2.3 kg.[2] The placebo group put on weight and its body-mass index went essentially nowhere, because its members also got taller. Body-mass index divides weight by the square of height, and in adolescents the denominator is still moving.
So the treated arm’s body-weight result, −14.8%, is close to what the same drug at the same dose did in adults over the same 68 weeks: −14.9% against −2.4%, a difference of 12.4 percentage points (95% CI, −13.4 to −11.5).[3] The adult and adolescent treated arms lost almost identical fractions of body weight. The gaps differ — 12.4 points in adults, 16.7 points of body-mass index here — because the control arms went opposite directions and because the two figures are not the same quantity. The cross-age comparison and the pharmacokinetic work behind it belong to the adolescents article; what matters on this page is that STEP TEENS reported both rulers and only preregistered one.
Two further adolescent-specific rulers appear in the registry. Body-mass index as a percentage of the 95th percentile fell 24.9 points against 4.5, and the body-mass index standard-deviation score fell 1.1 against 0.1.[2] Both are constructed to handle the growing denominator, and both move in the same direction as the headline.
How deep the responses went
The secondary confirmatory endpoint, a weight reduction of at least 5%, was reached by 73% of the semaglutide group against 18% of the placebo group.[1] Below that line the registry posts the deeper thresholds: at least 10% in 61.8% against 8.1%, at least 15% in 53.4% against 4.8%, and at least 20% in 37.4% against 3.2%.[2]
More than half the treated adolescents lost at least a seventh of their body weight. Waist circumference fell 12.7 cm against 0.5, and absolute body-mass index fell 5.9 units against 0.0.[2] On weight-related quality of life, measured with the IWQOL-Kids instrument, the total score and the physical-comfort domain improved on semaglutide under the primary estimand, and the total-score improvement was driven mainly by physical comfort; no other domain separated under either estimand.[1]
Seven weeks after the last injection
The trial built in a 7-week follow-up period with no injections, running to week 75, and reported the body-mass index change there as an exploratory endpoint. Body-mass index was 13.2% below baseline in the semaglutide group and 1.2% above baseline on placebo, an estimated difference of 14.4 percentage points (95% CI, −17.8 to −11.0), with lifestyle intervention still running in both groups.[1]
Read that carefully. About three points of body-mass index returned in seven weeks off the drug. The trial’s own discussion names the limitation: a longer follow-up period would have allowed the effect of treatment cessation to be monitored, given the small regain between weeks 68 and 75.[1] Seven weeks is not an off-treatment extension and this figure is exploratory and outside the testing hierarchy. What the adult withdrawal evidence looks like over a much longer horizon is in the discontinuation article.
The safety table
Adverse events were reported by 105 of 133 participants on semaglutide (79%) and 55 of 67 on placebo (82%) — a higher share on placebo — but the event rate ran the other way, 435.7 against 362.9 events per 100 person-years. Gastrointestinal disorders occurred in 62% against 42%, were generally mild or moderate, lasted a median of two to three days for nausea, vomiting and diarrhea, and peaked during or shortly after the 16-week escalation period.[1]
The individual counts are heavier than the adult tables. Nausea affected 56 of 133 (42%) against 12 of 67 (18%), vomiting 48 of 133 (36%) against seven of 67 (10%), diarrhea 29 against 13, headache 22 against 11, dizziness 10 against two.[2] Vomiting in more than a third of treated adolescents is the figure a parent would want stated first. What the symptoms are and how they are managed is in the side effects article.
Serious adverse events occurred in 15 of 133 (11%) against six of 67 (9%), with no deaths. A similar share of each group stopped the trial regimen because of adverse events — 5% against 4% — with gastrointestinal events the most common reason on semaglutide.[1] For scale, the adult pivotal trial recorded 4.5% against 0.8% discontinuing specifically for gastrointestinal events,[3] which is a narrower definition than the all-cause figure above and should not be subtracted from it.
Five participants (4%) on semaglutide had acute gallbladder disease, all five with cholelithiasis and one with concurrent cholecystitis; none on placebo did. Amylase and lipase rose on semaglutide, with no cases of pancreatitis. Heart rate rose a mean of 1.2 beats per minute on semaglutide and fell 2.3 on placebo. No clinically relevant findings were noted in biochemical, hematologic or growth measures, or in Tanner staging — recorded as safety observations, not reported as endpoints.[1]
The mental-health result, and who was not in the trial
Mental health was assessed with the PHQ-9 depression scale and the Columbia–Suicide Severity Rating Scale as an exploratory safety endpoint. Neither indicated a difference between groups, and psychiatric adverse events were reported by a smaller share of the semaglutide group than the placebo group, 7% against 15%.[1]
That reads reassuring, and it has to be read alongside the exclusion criteria. STEP TEENS excluded anyone with a major depressive disorder in the two years before screening, anyone with a diagnosis of a severe psychiatric disorder or bulimia nervosa, and anyone with a history of a suicide attempt.[1] The adolescents at highest psychiatric risk were not enrolled. An exploratory null result in a screened-clean cohort of 201 is not evidence that the drug is safe for the adolescents the trial screened out. What is known about this class and mood in adults is in the mental health article.
What STEP TEENS was not powered to show
It ran no active comparator: no other drug, no surgical arm, no intensive-lifestyle-only arm beyond the therapy both groups received.
It ran for 68 weeks with a 7-week tail. Final adult height, peak bone mass, pubertal trajectory and anything else that resolves over a decade are outside its horizon entirely, and 201 participants could not have powered a growth endpoint even if one had been defined. It enrolled nobody under 12. It enrolled 79% White participants and 62% girls, so subgroup questions by race or sex have no answer here. It carried no cardiovascular, renal or mortality endpoint. Its quality-of-life, week-75 and mental-health readouts are exploratory or unpowered by the trial’s own statement, and it did not test whether a second course works after a first one is stopped.
What an adolescent indication means, and what a reader here should do with it
An adolescent indication means a regulator has reviewed a trial like this one and permitted a prescriber to use the drug in that age group under specified conditions. It does not mean the drug has been shown safe across adolescence: the randomized exposure underneath that permission is the 68 weeks above, plus seven weeks of watching. It does not transfer to children younger than the enrolled range. And it is not a dosing instruction: the licensed adolescent maintenance dose and the age floors differ by product, and those specifics belong to the adolescents article and to the prescribing information, not to a trial report.
For this site’s readers the practical point is narrow and worth stating without hedging. Everything above happened inside a pediatric protocol: a 12-week supervised run-in, parents or guardians in the behavioral program, a 16-week titration a clinician could slow down, psychiatric screening at entry, and monitoring of growth, puberty, amylase, lipase and mood throughout. Treating a minor for obesity is a pediatric clinical decision made with a pediatric clinician. It is not something to arrange through an online seller, and the sellers listed on the compounded semaglutide board dispense compounded preparations, which are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they reach a patient. Nothing in STEP TEENS is a reason to buy a vial for a fifteen-year-old.
What an adult reader can legitimately take from this trial is narrower still: at the same dose over the same 68 weeks, adolescents and adults lost almost the same fraction of body weight, and the adolescent gastrointestinal burden was heavier. The famous 16% belongs to a ratio, not to a scale.