Every obesity drug trial runs a lifestyle program in both arms, and most of them run a thin one. STEP 3 ran a thick one deliberately: an eight-week low-calorie diet followed by thirty counseling visits across sixty-eight weeks, given to the placebo group and the semaglutide group alike. The question it was built to settle is whether the best available behavioral treatment and the drug add to each other. What it produced instead is the most awkward pair of numbers in the program — the largest weight reduction and the smallest measured drug effect — and the reason those two facts sit together is the whole finding. The trial it should be read against is STEP 1, which used a far lighter lifestyle intervention.
What was actually run
STEP 3 was a randomized, double-blind, parallel-group, 68-week phase 3a study at 41 sites in the United States, conducted from August 2018 to April 2020 in adults without diabetes who had either a body-mass index of 30 or greater, or 27 or greater plus at least one comorbidity. A total of 611 participants were randomized 2:1 to semaglutide 2.4 mg (n = 407) or placebo (n = 204). Both groups received a low-calorie diet for the first 8 weeks and intensive behavioral therapy — thirty counseling visits — across the full 68 weeks.[1]
The cohort was 495 women (81.0%), with a mean age of 46 years (SD 13), a mean body weight of 105.8 kg (SD 22.9) and a mean body-mass index of 38.0 (SD 6.7). Of the 611 randomized, 567 (92.8%) completed the trial and 505 (82.7%) were still receiving treatment at the end of it.[1] Roughly one participant in six had stopped the injections by week 68 while remaining in the trial.
The coprimary endpoints, and the placebo arm underneath them
The coprimary endpoints were the percentage change in body weight and the loss of 5% or more of baseline weight by week 68. At week 68 the estimated mean body-weight change was −16.0% with semaglutide against −5.7% with placebo, a difference of −10.3 percentage points (95% CI, −12.0 to −8.6; P < .001).[1]
The threshold outcomes followed. A loss of at least 5% was reached by 86.6% on semaglutide against 47.6% on placebo; at least 10% by 75.3% against 27.0%; at least 15% by 55.8% against 13.2% (P < .001 for all).[1]
Read the placebo column on its own. Nearly half of the people who received no active drug crossed the 5% threshold that regulators treat as a clinically meaningful response, and more than one in four crossed 10%. Those participants were doing the diet, the activity target and the counseling described in the diet article and the exercise article, and nothing else.
The comparison that inverts the selling point
STEP 1 ran the same molecule at the same dose for the same 68 weeks in 1,961 adults without diabetes, with a lighter lifestyle intervention. Its result was −14.9% against −2.4% on placebo, an estimated treatment difference of −12.4 percentage points (95% CI, −13.4 to −11.5).[2]
Line the two up. The semaglutide arm gained 1.1 points when the intensive program was added. The placebo arm gained 3.3. The measured effect of the drug therefore fell by 2.1 points, from −12.4 to −10.3, and the confidence intervals only just touch. The trial carrying the biggest absolute number is the trial in which the drug accounted for the smallest share of it.
That is a comparison between two trials rather than a randomized contrast, and it deserves the usual caution: separate cohorts, separate sites, separate years, nobody assigned to both protocols. What it is not is a stray result, because the direction has been tested directly.
Randomized evidence for the same direction
A 2019 trial assigned 150 adults with obesity to intensive behavioral therapy alone, to that therapy plus liraglutide 3.0 mg, or to a multicomponent arm adding a 1,200 kcal/day meal-replacement diet. Ninety-one percent completed one year. Mean weight losses were 6.1 ± 1.3%, 11.5 ± 1.3% and 11.8 ± 1.3% of baseline weight, with 44.0%, 70.0% and 74.0% losing at least 5%.[3]
The behavioral arm alone landed at 6.1%, within rounding distance of STEP 3's placebo arm at 5.7%. Two different sponsors, two different drugs, two different decades of protocol, and the counseling produced the same result both times. The meal replacements added 0.3 points on top of the drug, which is a reminder of how quickly additions stop adding.
A second trial ran STEP 3's shape in primary care. SCALE IBT randomized 282 participants to liraglutide 3.0 mg (n = 142) or placebo (n = 140), with the same Medicare-based behavioral benefit in both arms for 56 weeks. Weight loss was 7.5% against 4.0%, an estimated treatment difference of −3.4% (95% CI, −5.3 to −1.6; P = 0.0003), with 61.5% against 38.8% reaching 5%.[4] Once the counseling is in both arms, the drug column shrinks and the placebo column grows.
Thirty visits is not a normal lifestyle arm
A 2026 review synthesized the lifestyle components of phase 3 trials of FDA-approved obesity medications, covering 50 reports from 42 studies. Most set a calorie-restriction goal of 500 kcal/day and a physical activity goal of at least 150 minutes a week. Twenty-two of those studies provided counseling at least monthly, and a further 16 provided it at least every three months.[5]
Thirty visits across 68 weeks is roughly one every two and a half weeks, which sits at the far end of that distribution rather than in the middle of it. The design summary for the STEP program describes the same structural point: the trials differ from one another in the intensity of what surrounds the injection, not only in the population they enrolled.[6]
No telehealth weight-loss subscription supplies thirty counseling visits. A reader budgeting against this trial is buying one column of it, and the cost of the column that is actually for sale is what the cost calculator estimates.
The diet was front-loaded, and the drug was not
The low-calorie diet ran for the first 8 weeks only, while the counseling ran all 68.[1] For roughly the first eighth of the trial, every participant was doing the single most calorie-restrictive thing the protocol asked of them, and the semaglutide group was still escalating toward 2.4 mg rather than holding it.
That ordering matters for anyone reading an early weight curve as a forecast. The steepest part of the descent in both arms sits on top of an eight-week diet that then stopped, and the 2:1 allocation — 407 against 204 — means the placebo estimate rests on the smaller half of the trial throughout. A participant who is eight weeks into a telehealth prescription is not eight weeks into this protocol.
What ended treatment
Gastrointestinal adverse events were more frequent with semaglutide (82.8%) than with placebo (63.2%). Treatment was discontinued because of those events in 3.4% of semaglutide participants and 0% of placebo participants.[1]
Two readings sit side by side there. A zero in the placebo column means the symptoms that ended treatment were attributable to the drug rather than to the diet. And 3.4% is a small number, recorded in a trial where a study team managed escalation and a participant could raise a problem at a scheduled visit rather than by email. The dose ladder that produced it is described in the titration article.
What STEP 3 was never able to show
It had no arm that took semaglutide without the behavioral program, so the trial cannot say from inside itself what the counseling contributed. Every estimate of that contribution above comes from a different study.
It counted no cardiovascular events, measured nothing after week 68, and its own authors concluded that further research is needed to assess the durability of the findings.[1] The trial that carried semaglutide out to 104 weeks is described in the STEP 5 article, and what the weight curve does once treatment stops is in the stopping article. It also enrolled at United States sites only, in a cohort four-fifths female, with diabetes excluded.
What was in the syringe
Each figure here belongs to branded, FDA-approved semaglutide at a labeled dose, escalated under supervision, alongside a behavioral program that a buyer is not being offered. Most sellers on the semaglutide board dispense compounded preparations. Compounded drugs are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed, a distinction set out in the compounded-versus-brand article.
The practical consequence is a subtraction rather than an addition. A service quoting 16% is quoting a number produced by a drug plus thirty counseling sessions plus an eight-week supervised diet, in which the placebo column alone reached 5.7%. Working out which parts are included in a subscription is the first thing to do with the figure, and how that checking is done is described in the methodology.