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 [1]. The nutritional consequences of eating less are 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. In more familiar units that is about 271 kcal at one meal.
The pooled figure is also thinner than the paper's scope suggests. Sixteen studies entered the systematic review; 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).
What the analysis does establish is that the appetite effect is real, immediate and measurable rather than a matter of self-report — which matters because the alternative explanation, that people on these drugs simply report eating less, has never been fully ruled out by questionnaire studies. The mechanism behind it is described in nausea and appetite on one circuit, and the subjective version of the same experience in the effort and reward trial.
The authors close by arguing for structured nutritional counseling 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.