“It stopped working” is one of the most common things people say about these drugs after the first few months, and it is among the reasons people give for stopping altogether. This trial suggests that sentence may describe something real and mean something other than what it sounds like.
What was measured, and how
One hundred and twenty adults with overweight or obesity were randomized 3:2 to semaglutide 2.4 mg or placebo for 60 weeks. [1] At weeks 0, 20, 40 and 60 they came into a laboratory, rated their appetite while fasting and for four hours after a standardized breakfast, then ate as much as they wanted from a provided lunch. Food reward was assessed with a questionnaire about responsiveness to the pleasure of food.
Two different kinds of thing are being captured there. One is a measurement of behavior — how many calories somebody actually ate. The other is a report of an experience — how hungry they say they feel.
The behavior held
At the test lunch, the semaglutide group ate less than placebo by 291.9 kcal at week 20, 240.2 kcal at week 40, and 269.5 kcal at week 60, with standard errors of 64.4, 87.8 and 83.6. All three were significant. Fourteen months in, the gap was about where it started.
The feeling did not
At week 20, the semaglutide group reported significantly greater appetite suppression after the standardized breakfast, less hunger over the past week, and less preoccupation with food — differences of 2197, −17.4 and −14.9 on the respective scales, with standard errors of 823, 4.8 and 4.5.
At weeks 40 and 60, the groups did not differ significantly on any of the appetite measures. Food reward behaved in between: greater reductions than placebo at weeks 20 and 40, differences of −0.4 and −0.6. The abstract does not report a week-60 food reward result, so this page does not describe one.
What this changes about a common complaint
If somebody says the drug stopped suppressing their appetite at month ten, this trial suggests they may be accurately reporting their experience while being wrong about what it implies. The suppression they can feel is the difference from before. The suppression that matters is the difference from what they would otherwise eat, and only a comparison group can see that.
It is also a reason not to read side effects or sensations as a gauge of whether treatment is working, the same disconnect that showed up in the taste-function study and during the titration period.
What the trial cannot support
A laboratory lunch is not a diet. It measures what one person ate from one provided meal under observation, four times across fourteen months, and it is a proxy for daily intake rather than a record of it. People eat differently when they know they are being watched, in both directions.
The trial is also small for its ambition — 120 people split 3:2 is roughly seventy against forty-eight — and missing data were handled by imputation that assumes dropouts behaved like the placebo group, which is a deliberately conservative choice rather than a neutral one. And three of the five headline figures come without stated units, so they are reported here as scores and not converted into anything.
The part that matters for buying
This is the clearest mechanistic evidence that the effect is sustained while the drug is taken, which is a different claim from the effect being permanent. What happens to intake after stopping is a separate question with its own uncomfortable answer, and the arithmetic of a year is worth doing before the first order rather than at month ten — expected loss is an estimate, not a schedule.