Most reviews of this drug quote one number: about 15% of body weight, at 68 weeks. It is the right number and it is nearly useless to an individual reader, because the trial that produced it also published how widely the result varied — and almost nobody quotes that.
This page is built from the posted results of STEP 1 (NCT03548935), the FDA labels, and the adverse event reporting database, all read on 14 September 2026. It does not carry a score, a star rating or a verdict on whether anyone should take it.
What the trial actually reported
STEP 1 randomised 1,961 adults, 1,306 to semaglutide 2.4 mg weekly and 655 to placebo, for 68 weeks.
| Outcome at week 68 | Semaglutide 2.4 mg | Placebo |
|---|---|---|
| Mean change in body weight (in-trial) | −15.6% (SD 10.1) | −2.8% (SD 6.5) |
| Mean change (on-treatment) | −16.9% (SD 9.4) | −3.1% (SD 6.4) |
| Mean change in waist circumference | −14.1 cm (SD 9.6) | −4.4 cm (SD 6.9) |
The standard deviation is the part worth stopping on. 10.1 percentage points, around a mean of 15.6. In plain terms, the spread of individual results is roughly two thirds the size of the average result. This is not a drug whose effect clusters tightly around its headline.
The distribution, which is what a review needs
The trial published the thresholds as well, and they say the same thing in a form a person can use. Of the 1,212 semaglutide participants with a week-68 measurement:
| Weight reduction reached | Semaglutide (n=1,212) | Placebo (n=577) |
|---|---|---|
| at least 5% | 1,047 — 86.4% | 182 — 31.5% |
| at least 10% | 838 — 69.1% | 69 — 12.0% |
| at least 15% | 612 — 50.5% | 28 — 4.9% |
| at least 20% | 388 — 32.0% | 10 — 1.7% |
| did not reach 5% | 165 — 13.6% | 395 — 68.5% |
Two rows carry the whole page.
Half the treated group reached 15% — the figure the mean sits near — which means half did not. And 165 people, about one in seven, took semaglutide 2.4 mg for sixty-eight weeks under trial supervision and did not lose 5% of their body weight.
That is the single most useful fact any review of this drug can offer, and it is in the registry record rather than in the abstract. It does not make the drug a poor one; the placebo column, where 68.5% failed to reach 5%, settles that. It makes the average a bad summary.
What it costs, in events rather than dollars
| Semaglutide | Placebo | |
|---|---|---|
| Any non-serious adverse event | 1,052 of 1,306 — 80.6% | 447 of 655 — 68.2% |
| Serious adverse event | 128 of 1,306 — 9.8% | 42 of 655 — 6.4% |
| Deaths | 1 | 1 |
| Completed the trial | 1,240 — 94.9% | 609 — 93.0% |
| Withdrew by own decision | 26 | 17 |
Four in five people on the drug reported at least one adverse event. So did two in three on placebo, which is what an eighteen-month trial looks like in any arm. The gap that matters is the serious-event line: 9.8% against 6.4%.
The completion figures cut against the drug's popular reputation. Under trial conditions — escalation schedule fixed, drug free, a study nurse reachable — 95% of the semaglutide arm finished, and participant-initiated withdrawal was in single-digit percentages. Whatever drives the real-world discontinuation this class is known for, STEP 1 did not capture it, and this page should not be read as evidence that it does not exist.
What the reporting database adds: the reports are about use, not effect
FDA's adverse event system holds 73,001 semaglutide reports. The most-reported terms, read on 14 September 2026:
| Term | Reports |
|---|---|
| Nausea | 10,674 |
| Off label use | 7,325 |
| Vomiting | 6,964 |
| Diarrhoea | 6,462 |
| Decreased appetite | 4,905 |
| Constipation | 4,724 |
| Weight decreased | 4,487 |
| Impaired gastric emptying | 3,558 |
| Product use in unapproved indication | 3,483 |
| Blood glucose increased | 3,303 |
| Wrong technique in product usage process | 3,016 |
| Inappropriate schedule of product administration | 2,446 |
Four of the twelve most-reported terms describe how the drug was used, not what it did. Together they total 16,270 — more than nausea, the leading symptom. Roughly half of all semaglutide reports (36,179 of 73,001, 49.6%) are flagged serious, and 80.3% originate in the United States.
This is the same shape our tirzepatide review found, where the single most-reported term was a dosing error rather than a symptom. Two molecules, two databases, the same signal: a meaningful fraction of what gets reported about these drugs is about the circumstances of their use — compounded supply, off-label indications, escalation schedules people improvised. The database cannot separate those cases, and neither can this page. It can say that they are numerous enough to outrank the drug's most famous side effect.
The problem with the question itself
"Semaglutide reviews" treats the molecule as a product. It is five trade names across four FDA applications and two routes: Ozempic injection, the shared Ozempic tablets and Rybelsus oral line, Wegovy injection in several presentations including Wegovy HD and Wegovy FlexTouch, and Wegovy tablets. Different indications, different dose ceilings, different prices, and — as the oral products' own label arithmetic shows — wildly different amounts of drug to reach a comparable result.
STEP 1 tested one of them: 2.4 mg weekly by injection, in adults with obesity or overweight without diabetes. Everything on this page is that product, in that population.
What this review will not give you
A score. A recommendation. A comparison to tirzepatide dressed as a verdict — the trial evidence on that is STEP 8 and SURMOUNT-5, and it is an evidence question rather than a review one. And any account of what it feels like, which no registry record contains and no honest page can supply from one.
What it gives instead is the distribution. If a reader takes one thing from it, the useful thing is that the published spread around the average is wide enough that the average predicts very little about any individual, and that the trial said so itself.
Sources
ClinicalTrials.gov posted results for NCT03548935 (STEP 1) — participant flow, outcome measures with standard deviations and threshold counts, and the adverse events module — read 14 September 2026 through the registry's v2 interface. FDA prescribing information for Wegovy (effective 2024-04-23), Ozempic injection (2026-07-30) and the Rybelsus/Ozempic tablets label (2026-01-30), and FDA's adverse event reporting system, all via openFDA on the same day. Product and application counts from the FDA Orange Book data files dated 11 September 2026.