Give someone on a GLP-1 weight-loss drug a plate of food and tell them to eat as much as they want, and they stop about 270 calories short of where a placebo group stops. That is the pooled result of a meta-analysis ↗ published online July 25 in Diabetes and Metabolic Syndrome: Clinical Research and Reviews, the first to combine the small set of studies that actually measured food intake instead of asking people to remember it.
The authors searched six databases through March 2026. They kept randomized, placebo-controlled trials that ran at least four weeks and fed participants a standardized ad libitum lunch, meaning an unrestricted meal the study weighed before and after. Sixteen studies fit the broad review. Only three trials (four treatment arms, 209 people in total) reported the lunch measurement cleanly enough to pool. The pooled reduction was 1,132 kilojoules, or about 271 calories, with a confidence interval running from roughly 815 to 1,449 kilojoules. A kilojoule is just the metric cousin of the calorie, so this is a couple of hundred calories left uneaten at a single sitting. Unusually, the trials showed zero statistical heterogeneity, meaning they agreed with each other almost perfectly rather than scattering.
The two drugs in the pool were semaglutide, sold as Ozempic and Wegovy, and tirzepatide, sold as Mounjaro and Zepbound, which hits a second gut-hormone receptor (GIP) on top of the GLP-1 one. Tirzepatide produces more total weight loss in head-to-head trials, so the interesting null here is that the two were statistically indistinguishable on how much they cut a single meal (p = 0.15). Whatever gives tirzepatide its edge over a longer course of treatment, a bigger dent in one lunch is not obviously it.
The cleaner the meal test, the thinner the real-world claim. Adding an exploratory early-phase tirzepatide trial pushed the reduction up to 1,421 kilojoules. But it blew the agreement apart, with heterogeneity jumping to 70 percent. That is a sign the effect size is sensitive to exactly which studies you count. And the lunch is a lab artifact. It captures appetite on a single controlled afternoon, not what a person eats across a real week. The authors are blunt that habitual, free-living intake on these drugs is barely measured at all, which is a strange hole under the best-selling drug class of the decade.
Their practical recommendation is not about the drug but about what fills the smaller appetite: structured nutritional counseling that protects protein, because a shrinking plate that loses protein first is how a patient loses muscle along with fat.
The number puts a floor under a claim people usually make loosely. The appetite suppression is real and measurable, and it traces to specific wiring. A recent map of the GIP and GLP-1 systems located the appetite-suppressing signal in the brain's area postrema and hypothalamus ↗. What this meta-analysis adds is the size of the behavioral output those circuits produce at the table. peptidemodel hosts cards for both drugs measured here, semaglutide ↗ and tirzepatide ↗, each built on the GLP-1 receptor ↗ that carries the signal.