STEP (Semaglutide Treatment Effect in People with obesity) is the programme that produced the Wegovy label. Each trial answered a narrower question than "does it work". This page gives the primary result of each, as the paper reports it, and nothing the paper does not.
The table
Weight change is the estimated mean percentage change from baseline; "difference" is the estimated treatment difference versus placebo (or versus the comparator where stated) with its 95% confidence interval.
| Trial | Population | N | Weeks | Semaglutide 2.4 mg | Comparator | Difference (95% CI) | PubMed |
|---|---|---|---|---|---|---|---|
| STEP 1 | Adults, no diabetes | 1,961 | 68 | -14.9% | Placebo -2.4% | -12.4 (-13.4 to -11.5) | 33567185 |
| STEP 2 | Adults, type 2 diabetes | 1,210 | 68 | -9.6% | Placebo -3.4% | -6.2 (-7.3 to -5.2) | 33667417 |
| STEP 3 | Adults, intensive behavioural therapy | 611 | 68 | -16.0% | Placebo -5.7% | -10.3 (-12.0 to -8.6) | 33625476 |
| STEP 4 | Adults after 20-week run-in on drug | 803 | 48 (weeks 20 to 68) | -7.9% | Switched to placebo +6.9% | -14.8 (-16.0 to -13.5) | 33755728 |
| STEP 5 | Adults, no diabetes | 304 | 104 | -15.2% | Placebo -2.6% | -12.6 (-15.3 to -9.8) | 36216945 |
| STEP 6 | East Asia (Japan, South Korea) | 401 | 68 | -13.2% (1.7 mg: -9.6%) | Placebo -2.1% | -11.1 (-12.9 to -9.2) | 35131037 |
| STEP 7 | Predominantly East Asian (China and others) | 375 | 44 | -12.1% | Placebo -3.6% | -8.5 (-10.2 to -6.8) | 38330988 |
| STEP 8 | Adults, no diabetes, vs liraglutide | 338 | 68 | -15.8% | Liraglutide 3 mg -6.4% (placebo -1.9%) | -9.4 (-12.0 to -6.8) | 35015037 |
| STEP 9 | Obesity with knee osteoarthritis | 407 | 68 | -13.7% | Placebo -3.2% | reported as P less than 0.001 | 39476339 |
| STEP 10 | Obesity with prediabetes | 207 | 52 | -13.9% | Placebo -2.7% | -11.2 (-13.0 to -9.4) | 39089293 |
| STEP UP | Adults, 7.2 mg vs 2.4 mg | 1,407 | 72 | 7.2 mg: -18.7% | 2.4 mg -15.6%; placebo -3.9% | -3.1 (-4.7 to -1.6) vs 2.4 mg | 40961952 |
Reading across the rows
Thresholds. In STEP 1, 86.4% of semaglutide participants lost at least 5% of body weight (placebo 31.5%), 69.1% at least 10% (12.0%) and 50.5% at least 15% (4.9%). In STEP 2 the 5% figure was 68.8% versus 28.5%.
Duration. STEP 5 ran two years and landed within a point of STEP 1's 68-week result (15.2% versus 14.9%), which is the evidence that the effect holds while treatment continues. STEP 4 is the mirror image: people who reached 2.4 mg and then switched to placebo regained 6.9% over the next 48 weeks while those who continued lost a further 7.9%. The stopping guide is built on that trial.
Behavioural support. STEP 3 added 30 counselling visits and a low-calorie diet for 8 weeks. Placebo did better (5.7%) and so did semaglutide (16.0%), but the gap between them narrowed slightly compared with STEP 1. Support helps everyone; it does not replace the drug's effect.
Population. STEP 6 and 7 were run in East Asian populations with lower BMI entry criteria and showed slightly smaller percentage losses. STEP 6 is also the only trial with a randomised 1.7 mg arm, useful for the dose comparison. STEP 9 measured knee pain: WOMAC pain score fell 41.7 points versus 27.5. STEP 10 measured glucose: 81% of the semaglutide group reverted to normoglycaemia versus 14%.
Comparator. STEP 8 is the only head-to-head: semaglutide 2.4 mg weekly against liraglutide 3 mg daily, 15.8% versus 6.4%. Discontinuation for any reason was 13.5% versus 27.6%.
Tolerability in the same papers
Gastrointestinal adverse events were the common thread: 63.5% on 2.4 mg in STEP 2 (34.3% placebo), 82.8% in STEP 3 (63.2%), 82.2% in STEP 5 (53.9%), 84.1% in STEP 8 (82.7% on liraglutide), 70.8% on 7.2 mg in STEP UP (61.2% on 2.4 mg, 42.8% placebo). Discontinuation for gastrointestinal events was 4.5% versus 0.8% in STEP 1 and 3.4% versus 0% in STEP 3. The side-effect timeline turns the label's pooled table into per-symptom differences.
What the table does not tell you
Nothing here is about compounded semaglutide, which was in none of these trials. Nothing here is about a dose below 2.4 mg except STEP 6's 1.7 mg arm. And nothing here is a forecast for one person; the weight-loss projection tool on the FormBlends Calculators site applies these means to a body weight with that caveat printed on it. For plain-English digests of each trial's design and limitations, the FormBlends Research site covers STEP alongside SURMOUNT, SELECT and the rest, and FormBlends' own semaglutide research report reviews the programme as a whole.
Questions people ask
Why is the diabetes result so much smaller?
It is a consistent finding across GLP-1 trials. STEP 2 (type 2 diabetes) showed 9.6% loss versus 14.9% in STEP 1 (no diabetes) with the same dose and duration. The papers do not settle why; glucose-lowering co-medication, baseline metabolism and trial population all differ.
Are these the numbers I should expect?
They are population means from people who met trial criteria, attended frequent visits and received lifestyle counselling. Half of people did better and half did worse. Your prescriber's expectation for you should be built from your own response at each step, not from a table.
Sources
- Wilding JPH et al. STEP 1. N Engl J Med 2021. PubMed 33567185 Accessed September 4, 2026.
- Davies M et al. STEP 2. Lancet 2021. PubMed 33667417 Accessed September 4, 2026.
- Wadden TA et al. STEP 3. JAMA 2021. PubMed 33625476 Accessed September 4, 2026.
- Rubino D et al. STEP 4. JAMA 2021. PubMed 33755728 Accessed September 4, 2026.
- Garvey WT et al. STEP 5. Nat Med 2022. PubMed 36216945 Accessed September 4, 2026.
- Kadowaki T et al. STEP 6. Lancet Diabetes Endocrinol 2022. PubMed 35131037 Accessed September 4, 2026.
- Mu Y et al. STEP 7. Lancet Diabetes Endocrinol 2024. PubMed 38330988 Accessed September 4, 2026.
- Rubino DM et al. STEP 8. JAMA 2022. PubMed 35015037 Accessed September 4, 2026.
- Bliddal H et al. STEP 9. N Engl J Med 2024. PubMed 39476339 Accessed September 4, 2026.
- McGowan BM et al. STEP 10. Lancet Diabetes Endocrinol 2024. PubMed 39089293 Accessed September 4, 2026.
- STEP UP. Lancet Diabetes Endocrinol 2025. PubMed 40961952 Accessed September 4, 2026.
Canonical URL: https://formblendssemaglutide.com/guides/step-trials-by-the-numbers. Written by the FormBlends editorial team. This page is educational and is not medical advice; see the medical disclaimer.