Most side effects of this drug class run one way. Reflux does not. The mechanism that produces nausea should also produce heartburn: a stomach that holds a meal longer is a stomach under pressure for longer, and pressure at the top of the stomach is what opens the lower esophageal sphincter. Against that sits the single most reliable non-surgical treatment for reflux ever documented, which is losing a significant amount of weight. Both effects are real, they are measured in different studies, and they point in opposite directions. The general tolerability picture in what the trials recorded does not resolve which one dominates.
The class-level answer is a modest increase
The cleanest study design available for this question is an active-comparator new-user cohort, which compares people starting one drug against people starting a different drug for the same condition rather than against people starting nothing. A 2025 analysis of United Kingdom primary-care records did exactly that, following 24,708 new users of a GLP-1 receptor agonist against 89,096 new users of an SGLT-2 inhibitor, all with type 2 diabetes, for a median of three years.
Incident reflux disease was more common on the GLP-1: risk ratio 1.27 (95% CI, 1.14 to 1.42). Complications of reflux were also more common, at risk ratio 1.55 (95% CI, 1.12 to 2.29).[1]
The absolute figures are the half that rarely travels. The risk difference for reflux was 0.7 per 100 patients over three years, and for complications 0.8 per 1,000.[1]A 27% relative increase on a base rate that low is roughly one additional diagnosis for every 143 people treated for three years — a real effect and a small one.
The increase belongs to a different set of drugs
A larger study asked a sharper question. Using a global federated database, it built matched cohorts of 177,666 patients each and then split the GLP-1 arm by pharmacokinetics: agents with a half-life of a day or less against agents with a half-life of five days or more. Short-acting agents delay gastric emptying more than long-acting ones, which are subject to tachyphylaxis at the stomach.
Across the whole class, erosive reflux disease came in at hazard ratio 1.15 (95% CI, 1.09 to 1.22). Split by half-life, the entire excess sat on one side: short-acting agents at 1.215 (95% CI, 1.111 to 1.328), long-acting agents at 0.994 (95% CI, 0.924 to 1.069) — a confidence interval centered on no effect at all.[2]
The complications followed the same split. Esophageal stricture (HR 1.284), Barrett’s esophagus without dysplasia (1.372) and Barrett’s with dysplasia (1.505) were all elevated with short-acting agents and not with long-acting ones, and the pattern held when liraglutide, lixisenatide and exenatide were examined individually.[2]
That distinction decides who the finding is about. Semaglutide, tirzepatide and dulaglutide are once-weekly, long-acting drugs. Liraglutide is daily and exenatide and lixisenatide are shorter still. A page that publishes the 1.15 and stops has quoted a real number while describing the wrong molecule to almost everyone reading it, and the molecules being sold are the ones compared in the molecule article.
Esophageal function testing has not found a lesion
If these drugs disturbed the esophagus mechanically, high-resolution manometry would show it. A 2026 retrospective study compared patients with type 2 diabetes undergoing manometry on and off a GLP-1 receptor agonist, 132 patients in all, of whom 42 were users and 57.1% of those were on semaglutide.
Most users — 64.3% — had an entirely normal study. There was no significant difference in motility diagnoses between users and non-users (P = 0.07), no association between the specific drug or its dose and any diagnosis (P = 0.51), and among those who went on to formal reflux testing, no difference in detected reflux disease (P = 0.51).[3] The study is small and retrospective, and its authors frame the result as a reason not to interrupt treatment before esophageal testing rather than as proof of no effect.
The stomach end is where the change is visible
Endoscopy sees what manometry does not. A review of 1,094 patients undergoing upper endoscopy under deep sedation found increased residual gastric content in 20.33% of those who had taken semaglutide within 30 days, against 3.19% of those who had not (P < 0.001).[4]
That finding belongs to the anesthesia question and is handled there, in the surgery article. Its relevance here is narrower: a stomach with more in it is a stomach whose proximal wall is stretched more often, and stretch at the top of the stomach is the trigger for the transient sphincter relaxations that let gas and acid travel upward. It is the same reflex that produces the belching signal described in the eructation article.
The other direction is larger than the first
Weight loss does not nudge reflux; it resolves it in most people who achieve it. A prospective intervention enrolled 332 adults with a body-mass index between 25 and 39.9 in a structured program of diet, activity and behavior change. At six months, 97% had lost weight, an average of 13 kg.
Reflux prevalence fell from 37% to 15%(P < 0.01) and the mean symptom score from 5.5 to 1.8. Overall, 81% of participants had lower symptom scores; 65% had complete resolution and 15% partial, with a significant correlation between the percentage of body weight lost and the reduction in score.[5] Thirteen kilograms is well inside what the trials in the semaglutide results report.
A population cohort reached the same conclusion on a much larger scale. Nearly 30,000 Norwegians answered the same reflux questionnaire about a decade apart. Among those whose body-mass index fell by more than 3.5 units, the adjusted odds of losing any reflux symptoms were 1.98 (95% CI, 1.45 to 2.72) for people using no or less than weekly antireflux medication, and 3.95 (95% CI, 2.03 to 7.65) for people using it at least weekly. The relationship was dose-dependent: more weight lost, more symptom relief.[6]
Where the favorable half stops being true
The same analysis reports one result that does not fit the headline. For the loss of severe reflux symptoms, the odds ratio among people not on regular antireflux medication was 0.90 (95% CI, 0.32 to 2.55) — an interval straddling 1, meaning no demonstrated benefit. Among people who were taking antireflux medication at least weekly, severe symptoms did clear, at 3.11 (95% CI, 1.13 to 8.58).[6]
So weight loss reliably clears mild and moderate reflux, and its effect on severe reflux was only demonstrated alongside drug treatment. Anyone starting one of these prescriptions with established severe heartburn is in the subgroup where the reassuring result was not found, which is a conversation with a prescriber rather than a reason to stop an acid suppressant.
How the two forces sit on a calendar
Neither cohort study above could separate the drug’s mechanical effect from its weight effect, because both were conducted in type 2 diabetes where weight loss is more modest than in the obesity trials — the distinction drawn in the two-indications article. What the evidence supports is a sequence rather than a verdict: the mechanical effect is strongest during escalation, when the stomach holds more and the dose is climbing, and the weight effect accumulates over the following year.
Two limits close this out. Every figure here comes from studies of branded product, while most sellers reviewed on this site dispense compounded versions, which are not FDA-approved and are not reviewed by the FDA for safety, efficacy or quality before dispensing. And heartburn accompanied by difficulty or pain on swallowing, vomiting blood, or black stools is not ordinary reflux and warrants prompt medical contact rather than a change of diet — the same threshold that applies to the bowel symptoms in the constipation article.