Reduced Appetite on a GLP-1: the Label Never Lists It as a Side Effect, and the Three Trials Behind That Label Report It in 9% of People Against 3% on Placebo

Ask "how much will a GLP-1 take my appetite away?" and you are asking the one question about these drugs where the intended effect and the side effect are the same event. Every other entry in this series covers something nobody wants. This one covers the thing people are paying for — and that turns out to change where the number lives, and whether it is published at all.

This is the sixteenth piece in our evidence series, after which effects have enough evidence to calibrate a model, the corpus-derived pooled rates, what patients themselves report, gallstones end to end, what is inside a personalised risk number, hair loss end to end, why trial vomiting rates run from 6.5% to 58%, constipation read against the placebo arm, headache read against the placebo arm, abdominal pain, where the label and the registry disagree by a factor of two, nausea, where the label's 44% holds up and the largest trial reads 18%, diarrhoea, where the three trials behind the label's number add up exactly, fatigue, where one Phase 3 trial reports less of it on the drug, acid reflux and dizziness, where the two largest arms in the base have no placebo column at all. Like those, every figure below is quoted from a primary source with the arm and the number of participants it belongs to, and every one of them was read on 19 September 2026 — from the current US prescribing information, from each trial's own results table on ClinicalTrials.gov, or from our own live production API.

The label uses the word "appetite" three times, and never as an adverse reaction

The current US prescribing information for Wegovy (semaglutide), revised June 2026, uses the word "appetite" exactly three times. All three are in the pharmacology sections:

§12.1 Mechanism of Action — *"GLP-1 is a physiological regulator of appetite and caloric intake, and the GLP-1 receptor is present in several areas of the brain involved in appetite regulation."*

§12.1, continuing — *"Animal studies show that semaglutide distributed to and activated neurons in brain regions involved in regulation of food intake."*

§12.2 Pharmacodynamics — *"Semaglutide decreases calorie intake. The effects are likely mediated by affecting appetite."*

It appears in no adverse-reaction table, in no list of common side effects, and nowhere in the patient information. (Quoted from the prescribing information via DailyMed, set ID ee06186f, SPL version 19, effective 18 June 2026, accessed 19 September 2026.)

That is not an oversight, and this article is not an accusation. Reduced appetite is what the drug is for: a label that listed its own mechanism of action as an adverse reaction would be describing the treatment working. The sponsor's judgement here is defensible, and stated plainly in the pharmacology sections rather than hidden.

It has a consequence all the same. The document a reader, a clinician or a journalist would go to in order to find out how often people on this drug notice their appetite change does not carry the number — and the trials behind that document did record it.

Table 3's threshold, and the arithmetic that identifies its trials

The label's main adverse-reaction table for adults is explicit about what it includes: *"Table 3 shows adverse reactions reported in greater than or equal to 2% of WEGOVY 2.4 mg injection-treated adult patients and more frequently than in the placebo group from these trials."* Its columns are headed Placebo N=1,261 and WEGOVY Injection (2.4 mg Once Weekly) N=2,116, and it lists nineteen terms, from nausea at 44% against 16% down to dysesthesia at 2% against 1%.

Those two N values identify the pooled trials exactly, and the label's own footnote confirms one of them: footnote c defines the hypoglycaemia row using *"Study 3, WEGOVY N=403, Placebo N=402"*, arm sizes that match one trial in the programme and no other. Working from the registry records:

Both totals match the label's column headings exactly. (STEP 2 also ran a semaglutide 1.0 mg arm of 402 people; it is not part of the 2.4 mg pool, and leaving it out is what makes the arithmetic land.)

So we can ask a question the label leaves unanswered without stepping outside the label's own evidence base: what did those three trials post for decreased appetite?

What the three pooled trials actually reported

Every count below was read from that trial's own ClinicalTrials.gov adverse-events module on 19 September 2026. The counts are the registry's; the percentages, the sums and the differences are our arithmetic on them.

Rounded the way Table 3 rounds, that is 9% against 3% — clearing the table's stated threshold of at least 2% on the drug and more frequent than placebo, with room to spare. Ranked by the drug figure it would sit eighth or ninth of twenty rows, level with dyspepsia (9%) and above dizziness (8%). Ranked by the margin over placebo — the quantity that tells you how much to attribute to the drug — it would sit ninth: behind nausea (+28), vomiting (+18), diarrhoea (+14), constipation (+13), abdominal pain (+10), fatigue (+6), dyspepsia (+6) and eructation (+6 or more, its placebo figure being printed as "<1"), and ahead of headache (+4), hypoglycaemia in type 2 diabetes (+4), dizziness (+4), gastritis (+3) and every remaining row. Those margins are computed from the label's own rounded whole-percent figures, so they are accurate to about a point.

The three trials agree with each other unusually closely — 9.5%, 9.4% and 8.6% on the drug, 3.4%, 3.7% and 3.4% on placebo. That is a tighter spread than this series has found for any other effect at the same dose of the same molecule.

And the placebo columns deserve a second look on their own. Between 3.4% and 3.7% of people who spent 68 weeks injecting placebo reported decreased appetite as an adverse event — a reminder of what these rows are: a record of what participants volunteered to an investigator, not the output of a measuring instrument.

Beyond the label's three trials: 0.6% to 70%

Our database cited 23 distinct studies for a stated decreased-appetite rate when this article was written at 04:00 BST on 19 September 2026. Their per-study figures run across two orders of magnitude:

(Three further studies are in the base and omitted here only for length — a Phase 1 orforglipron ascending-dose study in 133 healthy subjects, a Phase 1 orforglipron study in 62 Japanese participants with type 2 diabetes, and a drug-interaction study of orforglipron with carbamazepine. All three appear on the live endpoint with their URLs.)

Four things in that table deserve to be pulled out.

The 70% is four people's worth of control, and half that control group reported it too. NCT06023095 is a 24-person Phase 1 titration study. Its two 10-person cohorts each posted 7 of 10, and its placebo arm posted 2 of 4 — 50%. The highest rate in our entire base comes from a trial in which half the people taking no drug at all reported the same thing. Quoted alone, "up to 70% of people on a GLP-1 lose their appetite" would be a true statement about one 10-person cohort and a useless one about anybody else.

One trial reports more decreased appetite on placebo than on the drug. PIONEER TEENS, in children and adolescents with type 2 diabetes, posts 2 of 66 on oral semaglutide against 4 of 66 on placebo. At those counts this is noise — and that is the point: at 66 people per arm, an effect the large trials measure at five to six points cannot be resolved at all.

Where the arms are large, the margin is remarkably stable. Take every placebo-controlled arm of more than 300 people — the four STEP arms, the three SURMOUNT-1 doses, the three ATTAIN-2 doses and SELECT, eleven arms in all — and the drug-minus-placebo difference runs from +3.5 to +12.4 points, with eight of the eleven between +5.2 and +6.2. The two extremes are dose-ordered in the direction you would expect: the smallest margin is STEP 2's lower 1.0 mg semaglutide arm (+3.5), the largest is ATTAIN-2's top 36 mg orforglipron dose (+12.4). Meanwhile the raw drug figure across those same eleven arms moves from 6.6% to 15.3%. The headline rate is unstable; the difference is not. That is the same pattern this series found for dizziness and fatigue, and it is why we keep printing the placebo column.

The rate does not climb cleanly with dose within a trial. SURMOUNT-1 posts 9.5% at 5 mg, 11.8% at 10 mg and 8.7% at 15 mg — the highest dose reports the least. ATTAIN-J posts 4.9%, 1.8% and 3.3% across its three ascending doses. ATTAIN-2 and ACHIEVE-J do rise monotonically. Whatever these rows are measuring, it is not a clean dose-response.

The trial designed to measure appetite reports the least of it

The sharpest illustration of what these numbers are is STABLE Wt Loss (NCT05548647), whose official title is *"Short- and Long-term Effects of Once Weekly Semaglutide 2.4 mg on Appetite, Eating Behavior, and Psychosocial Status"*. It is a Phase 4 trial whose entire purpose is measuring appetite.

Its adverse-event table posts decreased appetite for 4 of 72 people (5.6%) in the semaglutide-plus-behavioural-treatment arm, 0 of 48 on placebo, and 0 of 10 in the arm that continued semaglutide through the extension period.

A trial built to measure appetite reports appetite reduction as an adverse event less often than STEP 1 does, and in one arm not at all. Nothing is wrong with the trial. The explanation is that its appetite data are efficacy endpoints — measured on instruments, at scheduled visits, for every participant — while its adverse-event table is a different instrument entirely: a record of what participants spontaneously raised as a problem, coded to a standard dictionary. The same trial measured appetite carefully and recorded it as a complaint rarely, and only the second of those two numbers is the kind we are able to pool.

That is the honest limit on everything in this article, and on every registry-derived appetite figure anywhere. These rows do not measure how much appetite changes. They count how often somebody raised it, unprompted, as something wrong.

What our own number counts, and what a reader should know before quoting it

Our reduced-appetite endpoint publishes a pooled clinical estimate of 9.4% (95% CI 1–59%), from 123 eligible stated rates across 23 distinct studies, source diversity "high", as read at 04:00 BST on 19 September 2026.

Read the interval, not the point estimate. A 95% interval running from 1% to 59% is not precision with a caveat attached; it is the file telling you the studies genuinely disagree. They do — 0.6% in one Phase 1 orforglipron arm, 70% in another trial's 10-person cohort. Our 9.4% lands close to the 9.3% the label's own three trials pool to, and that closeness should be read as a coincidence rather than as corroboration: those three STEP trials are three of our twenty-three studies, and the rest of the pool spans five molecules, four phases and populations from healthy volunteers to people in a cardiovascular-outcomes trial.

And 123 rates is not 123 studies. Three Phase 1 orforglipron studies in healthy or otherwise-healthy volunteers supply 77 of the 123 rows — 63% (NCT06440980 alone posts 38 dose-level and formulation arms, NCT03929744 posts 21, NCT05841238 posts 18), because those studies move the same participants through cohort after cohort and period after period, and the registry posts each as its own arm. The three STEP trials, between them, supply four rows. Our pooling does not treat the 77 as independent observations — each study collapses to one entry, the unweighted mean of its posted non-placebo arms carried at its largest arm's size, and entries are then weighted by stated sample size — which is why the Phase 1 studies do not drag the estimate down towards 2%. But if you are reading the row count as evidence weight, read it as twenty-three studies, not one hundred and twenty-three findings.

What the FDA's adverse-event reports show

A different kind of number sits outside the incidence pool on purpose. FDA FAERS publishes the share of *spontaneous adverse-event reports* for each drug that mention decreased appetite, and our corpus holds four such rows: 4,790 of 82,377 semaglutide reports (5.8%), 2,113 of 43,057 liraglutide reports (4.9%), 3,153 of 101,618 dulaglutide reports (3.1%) and 5,643 of 94,688 exenatide reports (6.0%), as stored in our corpus and stated as of the dates those rows were collected.

A share of complaints is not a share of patients — it answers "of the people who filed a report about this drug, how many mentioned decreased appetite", not "how many people on the drug experienced it" — so our eligibility rules report these separately rather than averaging them into the pooled estimate. What is notable is that decreased appetite reaches the FAERS reports at all, in roughly the same share as it reaches the trial tables: people do file adverse-event reports about the effect the drug is prescribed to produce.

What patients themselves report

Of the 26 distinct community reports in our corpus, 8 mention reduced appetite — a 30.8% reporting frequency, the fourth-highest of the fifteen effects we track, behind vomiting (42.3%), nausea (38.5%) and diarrhoea (34.6%). That is not an incidence rate and we never treat it as one: it is the share of people who chose to post about their GLP-1 experience who mentioned appetite at all, and at eight reports it is a count, not a rate. See that article for why the pool is small and cannot currently grow.

The gap between that 31% and the trials' 9% is the most interesting number in this article, and we cannot close it. People writing about their own experience raise appetite change roughly three times as often as trial participants raise it with an investigator. Both are records of what somebody chose to mention; they differ in who was asked, why, and what counted as worth saying. Neither is a measurement of appetite.

If you are on a GLP-1 now

This is educational content, not medical advice. Reduced appetite is the intended action of these medicines and the label describes it as such. The trial evidence above says that when people on the drug raise it as a problem with an investigator, they do so about five to six points more often than people on placebo in the same trials — and that no source in our corpus states how *much* appetite falls, only how often someone flagged it.

What the label does warn about, in a related direction, is what can follow eating and drinking much less. It states that *"diarrhea, nausea, and vomiting may cause a loss of fluids (dehydration) which may cause kidney problems"*, and it names dizziness and light-headedness at the head of its list of low-blood-sugar symptoms, a risk it says is increased for *"patients with diabetes mellitus taking WEGOVY in combination with insulin or an insulin secretagogue"*. An appetite that has fallen far enough that you are not keeping fluids down, not eating enough to take a medication safely, or losing weight faster than you and your prescriber planned, is a reason to contact a clinician rather than something to treat as the treatment working. Nothing here should be used to start, stop or change a medication or its dose without your prescriber.

Every figure in this article was read on 19 September 2026 from the cited primary source or from our live production API, and is stated as of that date. Query everything behind it — all 15 effects, both tracks, every source with its URL — at magistra.health/en/data-api, with the full eligibility methodology at magistra.health/en/methodology.