Nausea is the effect people brace for, and it is also the one that keeps a schedule: it arrives with a dose change and settles while the dose holds, a shape laid out in the timing article. Constipation does not behave that way. It has no obvious relationship to the last injection, it can appear in month four as easily as week two, and the standard explanation for it does not survive the studies that measured the thing being explained.
What the randomized trials recorded
The largest synthesis restricted to the population that buys these drugs for weight rather than for diabetes pooled 39 reports covering 33,354 individuals with overweight or obesity and without diabetes. Nausea, vomiting, diarrhea and constipation were the four most common gastrointestinal events, and every agent examined raised the risk of nausea.[1]
Constipation did not follow the class. The increase reached significance for semaglutide and for liraglutide, and did not reach significance for cagrilintide or for exenatide.[1] That is the first sign that constipation is not simply what happens when a GLP-1 receptor is occupied, because four drugs occupying the same receptor did not produce the same bowel.
Dose moves the gastrointestinal total in the direction you would expect. STEP UP randomized 1,407 adults to semaglutide 7.2 mg, 2.4 mg or placebo, and gastrointestinal adverse events were reported by 70.8%, 61.2% and 42.8% of each group respectively.[2] The placebo figure is the part worth keeping: more than four in ten people on no drug at all reported a gastrointestinal event over 72 weeks, which is the baseline any personal experience has to be read against. What each schedule looks like is in the titration article.
The slowed-stomach explanation, measured directly
Delayed gastric emptying is the mechanism almost every consumer page offers, and it is real early in treatment. Whether it is still there at the maintenance dose is a different question, and one trial asked it. A double-blind study randomized 72 adults with obesity to semaglutide escalated to 2.4 mg or placebo for 20 weeks, and measured gastric emptying by paracetamol absorption after a standardized breakfast.
The five-hour absorption area was 8% higher on semaglutide (P = 0.005), which sounds like a delay until the correction lands: adjusted for week-20 body weight it was not significant (P = 0.12). Nothing moved at one hour, nothing moved in peak concentration, and nothing moved in time to peak. The authors state plainly that there was no evidence of delayed gastric emptying at week 20.[3]
That result has a genuine limit — paracetamol absorption is an indirect measure, and a scintigraphic study can find a delay where this method does not. But it is the measurement taken at the dose and the duration most people are actually on, and it points away from a stomach that stays slow for as long as the constipation does.
The colon was measured once, and it sped up
A slow stomach would not explain hard stool anyway; that is a colonic event. Exactly one study put a GLP-1 analog and validated scintigraphic colonic transit into the same protocol. Forty-six women with constipation-predominant irritable bowel syndrome received ROSE-010 at 30, 100 or 300 micrograms, or placebo.
Gastric emptying of solids was significantly retarded at the two higher doses — the expected finding. Colonic transit at 24 hours was not retarded at all, gastric accommodation did not change, and bowel function did not change. At 48 hours the 30- and 100-microgram doses accelerated colonic transit.[4] The paper's stated conclusion is that the drug showed potential for relief of constipation in that population.
So the one direct human look at the organ responsible found the opposite of the mechanism in circulation. That is a small study in a specific population using a compound nobody prescribes, and it does not close the question. It does mean the confident sentence about a drug slowing the whole digestive tract is an inference, not a measurement, and the broader tolerability picture in what the trials recorded never established it either.
What is left is how little goes in
The same 20-week trial that found no emptying delay found something large. At a free-choice lunch, energy intake was 35% lower on semaglutide than on placebo — 1,736 against 2,676 kilojoules, a treatment difference of 940 kilojoules with P < 0.0001. Hunger fell, fullness rose, and body weight dropped 9.9% against 0.4%.[3]
A third less food is a third less fiber and, for most people, meaningfully less fluid, because a large share of daily water arrives inside meals and alongside them. Constipation on this class looks far more like the downstream consequence of that than like a motility side effect, which reframes what to do about it: the lever is what goes in, and the same logic governs the protein question in the diet article.
Fiber has a number, and the number has a price
Fiber supplementation in chronic idiopathic constipation has been pooled. Across seven randomized trials, 113 of 147 patients assigned to fiber responded against 61 of 140 on placebo — 77% against 44%, a risk ratio of 1.71 (95% CI, 1.20 to 2.42; P = 0.003). Stool frequency rose and consistency softened, both significantly.[5]
The same analysis reports the cost in the same breath. Flatulence was significantly higher on fiber than on placebo (SMD 0.56, 95% CI 0.12 to 1.00; P = 0.01), and the authors graded the overall quality of evidence as low with a high risk of bias. Extra gas is not a neutral trade on a drug whose most disproportionately reported gastrointestinal event is belching, which is the subject of the eructation article.
Water is an adjunct, not a treatment
The advice to drink more is usually given on its own. The trial behind it did not test it that way. One hundred and seventeen adults with chronic functional constipation ate roughly 25 g of fiber a day for two months; one group drank freely, averaging 1.1 liters, and the other was instructed to drink two liters, averaging 2.1.
Both groups improved on fiber alone. The higher-fluid group improved more, on stool frequency and on laxative use, both at P < 0.001.[6] The finding is conditional: fluid amplified fiber. It was never shown to work without it, and no trial of these drugs has measured what happens to a patient’s fluid intake when their meals shrink by a third.
The guidance is mostly borrowed
A 2026 international panel ran a modified Delphi process and produced 52 consensus statements on nutrition and lifestyle support during treatment, covering nausea, vomiting, diarrhea and constipation. The statements were primarily derived from indirect evidence — existing guidelines for nutrition therapy in bariatric medicine, plus clinical experience — and the paper closes by saying direct evidence is urgently required.[7]
That is the honest status of nearly every bowel tip attached to a telehealth prescription. It is reasonable advice imported from an adjacent field, not a result from a trial of the drug you are taking.
Molecule and reporting instrument both matter
A pharmacovigilance study of 123,145 tirzepatide reports compared them against semaglutide reports and found tirzepatide carried lower reporting odds of vomiting, constipation and decreased appetite.[8] That is a spontaneous reporting database with no denominator and heavy channeling by prescribing fashion, so it cannot produce an incidence. It can still say that the two molecules are not interchangeable on this symptom, which the head-to-head differences in the molecule comparison support from the other direction.
Two boundaries close this out. Every figure above comes from trials or registries of branded product at labeled doses, while most sellers on this site dispense compounded versions, which are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before they are dispensed. And constipation that comes with severe abdominal pain, vomiting, or no bowel movement for several days is not the ordinary version of this and belongs with a prescriber rather than with a fiber supplement — the same threshold that applies to the higher rates reported in adults over 65.