About 271 calories at one observed meal, and nobody has measured a whole day properly. A systematic review pooled standardized test-meal data from four arms across three trials, 209 participants in total, and found energy intake fell by 1,132 kJ against placebo, on a confidence interval from −1,449 to −815 [1]. In more familiar units that is roughly 271 kcal at a single lunch. A separate 60-week trial in 120 adults measured the same thing four times and found the gap held: 291.9 kcal at week 20, 240.2 at week 40 and 269.5 at week 60 [2]. What faded over those months was the feeling of it, not the eating.
Everyone knows these drugs make people eat less. Almost nobody has measured how much, because asking people what they ate produces unreliable answers and weighing it is expensive. A systematic review searched six databases for trials that measured eating directly, under observation, and pooled what it found. The nutritional consequences of eating less are set out in vitamins and minerals on a GLP-1.
The pooled result is precise. At a standardized ad libitum lunch, participants on a GLP-1 or dual agonist ate 1,132 kJ less than on placebo, with a confidence interval from −1,449 to −815 and an I² of 0%, meaning the contributing trials agreed almost exactly.
The pooled figure is also thinner than the paper’s scope suggests. Sixteen studies entered the systematic review, and four arms from three trials, 209 participants in total, contributed to the meta-analysis. Adding an exploratory three-week phase 1 tirzepatide trial raised the estimate to −1,421 kJ and pushed heterogeneity from zero to 70%, which shows how much a single study can move a pool this size. No significant difference appeared between semaglutide and tirzepatide (P = 0.154).
Whether it lasts, and what stops lasting
The longest measurement of the same thing ran 60 weeks and randomized 120 adults 3:2 to semaglutide 2.4 mg or placebo [2]. At a laboratory lunch the semaglutide group ate 291.9 kcal less than placebo at week 20, 240.2 kcal less at week 40 and 269.5 kcal less at week 60, with standard errors of 64.4, 87.8 and 83.6. All three differences were significant, so fourteen months in, the gap was about where it started.
The self-reported measures did not hold. At week 20 the semaglutide group reported greater appetite suppression, less hunger over the past week and less preoccupation with food. At weeks 40 and 60 the groups did not differ significantly on any appetite measure at all. Why that combination is not a contradiction, and what it means for the common complaint that the drug stopped working, is in appetite at fourteen months.
The mechanism, and why it is hard to separate from nausea
What the analysis establishes is that the appetite effect is real, immediate and measurable rather than a matter of self-report. That matters because the alternative explanation, that people on these drugs simply report eating less, has never been ruled out by questionnaire studies. Work in mice traced the anorectic response and the aversive one to a single circuit in the dorsal vagal complex [4], which is a finding about rodents and a reason the two effects are hard to prise apart. It is described in nausea and appetite on one circuit.
What gets eaten, not only how much
A cross-sectional study gave a 53-item taste test and a 40-item smell test to 46 people on a GLP-1 and 46 matched controls [3]. Taste scores averaged 28.61 (95% CI 25.66 to 31.56) against 40.63 (38.35 to 42.91), at p<0.001 and an effect size of η²=0.37, while smell was slightly lower and not significantly different at p=0.076.
Nobody’s taste was measured before they started, so this shows that taste scores differ substantially rather than that the drug caused it. What makes it matter for intake is what people do when food registers as flat, which tends to be drifting toward salt, sugar and texture. The study and its one result that points the wrong way are in the taste-function study.
What follows from a smaller plate
The authors of the pooled analysis close by arguing for structured nutritional support with attention to adequate protein, which follows from their own finding. If intake drops by this much at a single meal, composition matters more than it did before, and nothing about the drug ensures that what remains is well chosen. That connects directly to the body composition question in what these drugs do to muscle.
For a buyer the useful translation is 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 has its own uncomfortable answer, and no seller on this roster publishes a nutrition plan to go with the prescription.