Metabolic & Cardiometabolic
GLP-1 receptor agonists causes GI side effects via delayed gastric emptying (nausea, vomiting; rare gastroparesis)
In plain terms: Do GLP-1 drugs commonly cause stomach and GI side effects?
Strong support Metabolic & Cardiometabolic 💰 Industry COI noted
Part of: 💊 GLP-1 Drugs
RefutedContestedStrong support
consensus score 0.93
Very commonly nausea, vomiting, diarrhea and constipation — usually mild/transient and tied to delayed gastric emptying — with a small but real increase in rarer events like gastroparesis and bowel obstruction.
📅 Last reviewed: 2026-07-15 ⓘ
Evidence ladder
How far up the ladder this claim has climbed. A high consensus on a low rung means "consistent so far," not "proven in people."
Top evidence so far: All trials, pooled (Meta-analysis)
MechanismIn-vitroAnimalObservationalRCTMeta-analysis
How the studies fall
16 support 0 contradict 0 tested null 0 mixed · 17 sources, 16 independent groups
The evidence (31)
| Source | Grade | Stance | Quality | Finding |
|---|---|---|---|---|
| Hjerpsted, Flint, Brooks, Axelsen, Kvist, Blundell 2018 · Diabetes Obes Metab | RCT | moderate | Semaglutide 1.0 mg weekly vs placebo, 12-week double-blind crossover in 30 adults with obesity (NCT02079870): 'gastric emptying during the first hour, following the standardized carbohydrate-rich breakfast, was 27% lower with semaglutide, compared with placebo (AUC 0-1h ETR: 0.73 [95% CI: 0.61, 0.87]; P = .0012)', and 'Two subjects withdrew because of gastrointestinal adverse events during the first treatment period while receiving semaglutide'. GI symptoms were NOT an endpoint — the paper reports no adverse-event table and never uses the words nausea, vomiting or satiety; the emptying delay is linked analytically only to the early glucose response (~40% of it). Tempering: at 12 weeks the overall emptying delay was absent (AUC 0-5h ETR 0.94 [0.88, 1.01]) and the authors attribute this to adaptation over time. Wholly Novo Nordisk funded; 4 of 6 authors are employee-shareholders. | |
| Zaitoon H et al 2025 · Children (Basel) | observational | supports | low | Narrative review of pediatric GLP-1RA use (structured search, 11 RCTs tabulated, 'No new data were created or analyzed in this study'): 'GI adverse events - most commonly nausea and vomiting - are the primary tolerability issue, while mean heart rate increases are minimal and treatment discontinuations are rare'; 'Nausea and vomiting are the most common adverse events, leading to discontinuation in 5-10% of adolescent trial participants'. In children 6 to <12 y, liraglutide lowered BMI vs placebo 'with gastrointestinal (GI) effects matching the expected safety profile'. Secondhand over the primary trials and meta-analyses it cites; authors declare Novo Nordisk trial-investigator ties. |
| Kim MK et al 2026 · Journal of Obesity & Metabolic Syndrome | observational | supports | moderate | Narrative review, delayed-gastric-emptying section: 'GLP-1 decreases gastric emptying by inhibiting stomach peristalsis while increasing tonic contraction of the pyloric region', with 'approximately 3.4% to 58.4% of GLP-1 RA users' reporting GI effects including nausea and vomiting, and 'more recent evidence demonstrates a significant relationship between GI symptoms and residual gastric content'. Cited cross-sectional data: '56% of participants using GLP-1 RAs had increased residual gastric content, compared with only 19% of non-users', and in a retrospective study 'approximately 10% of patients had retained gastric contents, the risk was three times higher among GLP-1 RA users than for non-users'. Secondary source: no primary data of its own, so weight is corroborative only. |
| Xie 2025 · Diabetes Metab Syndr Obes | meta-analysis | supports | moderate | Network MA, 19 RCTs / 13,529 patients with obesity or overweight, placebo in the network. Versus placebo (Figures 5A/5D/5G): nausea OR 4.43 (3.36,5.83) liraglutide 3.0 mg and 4.29 (3.45,5.34) semaglutide 2.4 mg; vomiting 4.39 (3.41,5.64) and 4.86 (3.84,6.15); diarrhea 1.92 (1.36,2.71) and 2.23 (1.68,2.97); constipation 2.14 (1.56,2.95) and 2.73 (2.04,3.66); dyspepsia 3.25 (2.37,4.46) and 2.97 (2.07,4.25); abdominal pain 2.29 (1.21,4.33) and 2.70 (1.61,4.53) - all significant. Between agents the GI burden is indistinguishable: 'There were no significant differences were observed in the incidence of diarrhea, nausea, vomiting, constipation, dizziness, abdominal pain, dyspepsia, upper respiratory tract infection, decreased appetite, headache, and nasopharyngitis between the groups treated with tirzepatide, semaglutide, and liraglutide.' |
| Sekhon S, Kahlon I, Latson W 2026 · Cureus | observational | supports | low | OFF-SCOPE for this claim (wrong outcome). Single-patient case report, n=1, Cureus, full text read. The exposure fits the class - 'Six weeks before admission, she had begun weekly tirzepatide injections for weight reduction' - but the measured outcome does not fit the claim: 'Colonoscopy identified patchy erythema, superficial ulcerations, and erosions extending through the sigmoid and descending colon, with normal findings in the rectum and terminal ileum', and 'Biopsies from the descending colon demonstrated crypt injury and ulceration in keeping with ischemic colitis.' Nausea, vomiting and gastroparesis - the outcomes this claim names - were never recorded, and no gastric-emptying measurement was performed. The paper itself keeps the two buckets apart: 'while gastrointestinal side effects are commonly seen, ischemic colitis has been only rarely described', and the sentence naming the claim's mechanism is background citing another author - 'Wichelmann et al. suggested that delayed gastric emptying and altered splanchnic perfusion could predispose to colonic hypoperfusion' - immediately followed by 'The exact mechanism still remains unknown.' The case is well worked up for its design (normotensive on arrival, 'blood pressure 118/82 mmHg, heart rate 74 beats per minute', so shock is excluded; stool studies negative; no NSAIDs or anticoagulants; graded 'moderate level 4' on the Hartwig and Siegel scale), but the authors state plainly that 'This report is limited by its single-patient design, which precludes establishing causality or generalizing findings to broader populations.' The finding is recorded as a candidate for a future 'GLP-1 receptor agonists cause ischemic colitis' claim; it is not evidence for this one. Prior extract, written from the abstract, generalised it to 'drug-induced GI complication' and thereby smuggled a different outcome onto this claim. |
| Witaszek T et al 2025 · Biomedicines | observational | supports | moderate | Narrative review (PubMed + Scopus, 2006-2025, last search 8 Sep 2025; no data of its own). 'Clinical trial data on AOMs indicate that GI AEs are reported in 65-84% of patients treated with liraglutide, semaglutide or tirzepatide'; 'The most common GI AEs across all AOMs are nausea and diarrhea' (STEP-8 nausea 61.1% [95% CI 52.4-69.2], constipation 39% [30.8-47.6] on semaglutide 2.4 mg; diarrhea 18-36% across incretin AOMs). The review qualifies the mechanism route: gastric-emptying delay is only one of several central and peripheral mechanisms, 'Most studies examining the relationship between GLP-1RA-induced gastric emptying delay and GI symptoms in obesity have found only weak or no correlations', and tachyphylaxis to the emptying effect means 'persistent GI symptoms cannot be explained by this mechanism alone'. |
| Newsome 2021 · N Engl J Med | RCT | supports | high | MASH trial: nausea 42% vs 11%, vomiting 15% vs 2%, constipation 22% vs 12% with semaglutide — reproduces GI signal across indications. |
| Xie X et al 2025 · Frontiers in Pharmacology | meta-analysis | supports | moderate | Bayesian network meta-analysis, 48 RCTs / 27,729 adults with T2DM (PROSPERO CRD42024592308, no funding, no COI declared). Placebo comparison — the limb this claim rests on — is uniform: 'while other GLP-1RAs significantly increased the incidence of nausea' and 'Compared to placebo, all GLP-1RAs significantly increased the risk of dyspepsia'. Pooled incidence: overall GI AEs 11.66%; 'nausea (21.49%), diarrhea (10.62%), vomiting (9.10%), dyspepsia (8.67%), constipation (7.92%), and decreased appetite (5.49%)'. Quality moderate, not high: the paper contradicts itself repeatedly on the between-drug ranking — the Discussion says 'A total of 45 randomized controlled trials (RCTs) were included, involving 27,729 patients' against the Results' 48; the abstract's 'dulaglutide and exenatide showed relatively better tolerability' is refuted by its own body, 'The highest incidence of nausea was observed with exenatide, at 31.70%'; the abstract puts vomiting highest with exenatide while section 3.6 says 'Lixisenatide had the highest risk of inducing vomiting, while dulaglutide had the lowest'; and it reports a point estimate outside its own interval, 'neither tirzepatide (OR = 2.21, 95% CI: 0.401–2.10) nor lixisenatide (OR = 0.48, 95% CI: 0.13–1.79) showed a significant difference in the risk of constipation'. 31% of included trials were at high risk of bias for allocation concealment and 18% for participant blinding, which matters because every outcome is a self-reported symptom. Supports the symptom limb of the claim only: the paper measures no gastric emptying and no gastroparesis, asserting the mechanism narratively — 'GLP-1RAs primarily influence gastrointestinal function by inhibiting gastric emptying and activating central GLP-1 receptors'. |
| Jones M et al] # corrected 2026-08-17 from PubMed; was [2025 2025 · Cureus | observational | supports | low | Case report/review: rare bowel obstruction with GLP-1RA; biologically plausible via motility slowing, but causality uncertain in case data. |
| Jalleh RJ et al 2026 · J Clin Invest | observational | supports | moderate | Narrative review (J Clin Invest 2026;136(4):e194740), no new data - zero-weighted. On this claim it is firmly supporting: 'A recent systematic review of 39 randomized controlled trials of GLP-1RAs showed a class effect of increased risks of nausea, vomiting, diarrhea and constipation compared with placebo' (nausea RR semaglutide 2.95, tirzepatide 2.90, orforglipron 4.77); in a second review of 38 phase III/IV placebo-controlled RCTs in T2D, 'nausea was reported in 19.3% of participants with active treatment versus 6.5% with placebo, and vomiting was reported in 7.6% of participants versus 2% with placebo'; discontinuation for adverse events 6.5% vs 3.6% placebo, nausea the leading cause. Delayed emptying is real and large (retained gastric contents 56% vs 19%) but the review does NOT attribute the symptoms to it: it offers a brainstem/area-postrema route as an alternative and states that 'The effects to slow gastric emptying, induce nausea, and reduce energy intake are independent' - so the claim's effect is well supported while its 'via delayed gastric emptying' mechanism is only partly so. Counterweight from the same paper: 'a substantial proportion (>50%) of the participants did not report any GI adverse events'. The authors also argue self-reported symptom capture is unstandardised and nocebo-prone, which weakens every rate quoted in the class. |
| Dahl, Brooks, Almazedi, Hoff, Boschini, Baekdal 2021 · Diabetes Obes Metab | RCT | supports | low | Double-blind crossover in 15 T2D subjects: oral semaglutide 14 mg produced 93 AEs in 14 subjects vs 51 in 13 on placebo, and 'the difference in the number of events was primarily due to gastrointestinal AEs with oral semaglutide'; the named mechanism was measured directly — paracetamol absorption 31% lower in the first hour (ETR 0.69; 95% CI 0.54-0.87; p = .0050) though unchanged over 5 h. Supports the claim, but low quality: n = 15, Novo Nordisk sponsored with 4/6 author-employees, and the authors state 'no correction for multiplicity was performed and so there is a risk of false positive results'; no nausea/vomiting counts appear in the text (supplementary Table S9 only). |
| Qian Z et al 2026 · PLoS One | observational | supports | low | FAERS 2004-2025 (>58M reports) + CVARD disproportionality signal detection, 7 MedDRA motility/reflux PTs across the top 50 drugs: all four GLP-1RAs were signal-positive on all three algorithms — 'semaglutide demonstrated the strongest association with impaired gastric emptying (ROR: 80.27, 95% CI: 76.39-84.34), followed by dulaglutide (ROR: 37.28, 95% CI: 34.90-39.83) and liraglutide (ROR: 10.82, 95% CI: 9.20-12.73)', with exenatide and semaglutide also elevated for gastric hypomotility (10.06 and 14.43); semaglutide replicated in CVARD (ROR 54.17, 95% CI 39.34-74.59) and onset was early (median TTO exenatide 7.6 d, liraglutide 9.0 d, semaglutide 28.8 d; Weibull beta <1 = declining hazard). Quality low, not moderate: this is denominator-free spontaneous reporting, and the authors state that 'the absence of a denominator (total number of patients exposed) precludes calculation of true incidence rates' and that ''stimulated reporting (notoriety bias)' - where increased media attention or regulatory actions stimulate increased reporting - may amplify signals for high-profile drugs such as GLP-1RAs', i.e. the dominant bias points in the direction of the recorded stance and is not corrected by the CVARD replication (the same worldwide semaglutide publicity inflates both databases). Their own data carry the negative control: metoclopramide, a prokinetic prescribed FOR gastroparesis, is itself 'signal-positive' for impaired gastric emptying (ROR 2.99, 95% CI 2.34-3.82), which the authors concede 'likely represents indication bias rather than true pharmacological toxicity'. Note also that nausea and vomiting are not among the seven PTs searched, so this source votes on the delayed-gastric-emptying limb of the claim only. |
| Wilding JPH, Batterham RL, Calanna S, Davies M, Kushner RF 2021 · N Engl J Med | RCT | supports | high | STEP 1 (NCT03548935): double-blind RCT, n=1961 adults with obesity and without diabetes, semaglutide 2.4 mg weekly vs placebo 2:1, 68 weeks. Abstract-verifiable GI finding: 'Nausea and diarrhea were the most common adverse events with semaglutide; they were typically transient and mild-to-moderate in severity and subsided with time', and 'More participants in the semaglutide group than in the placebo group discontinued treatment owing to gastrointestinal events (59 [4.5%] vs. 5 [0.8%])'. UNVERIFIED HERE: the per-symptom incidences previously recorded on this note (nausea 44%, diarrhoea 32%, vomiting 25%, constipation 24%) appear nowhere in the abstract, which is the only text this paywalled paper has ever been read from (provenance: abstract, no PMCID, isOpenAccess N) - confirm against the full paper's safety table before quoting them publicly. Novo Nordisk funded; three authors are Novo employees. |
| Kristensen SL, Rørth R, Jhund PS, Docherty KF, Sattar N, Preiss D, Køber L, Petrie MC, McMurray JJV 2019 · Lancet Diabetes Endocrinol 2019;7(10):776-785 | meta-analysis | supports | high | OFF-SCOPE for GI tolerability. This CVOT meta-analysis states precisely which harms it pooled: 'We did a meta-analysis using a random-effects model to estimate overall hazard ratios (HRs) for MACE, its components, death from any cause, hospital admission for heart failure, kidney outcomes, and key safety outcomes (severe hypoglycaemia, pancreatitis, and pancreatic cancer).' The corresponding result reports only those three — 'There was no increase in risk of severe hypoglycaemia, pancreatitis, or pancreatic cancer.' Nausea, vomiting, gastric emptying, gastroparesis and drug discontinuation for intolerance appear nowhere in the paper's outcomes. The previous extract ('CVOT MA notes GI AEs are the principal tolerability limitation of the class, exceeding placebo across trials') asserts a placebo comparison and a tolerability ranking that this paper never performed. Caveat kept honest: the full text is paywalled (Europe PMC: isOpenAccess N, 'Subscription required'), so a supplementary AE table cannot be positively excluded — but the METHODS sentence affirmatively enumerates the pooled safety outcomes, which is scope evidence rather than mere silence. |
| Hageen AW et al 2026 · Endocrinology, Diabetes & Metabolism | meta-analysis | supports | moderate | Network meta-analysis of 5 double-blind RCTs (n=3594; orforglipron 2385 / placebo 1209), PROSPERO CRD420251249505: 'All evaluated OFG doses (3, 12, 24, 36 and 45 mg) were associated with significantly increased odds of gastrointestinal adverse events compared with placebo', with a stated dose-response for nausea, vomiting, diarrhoea, constipation, dyspepsia and eructation at doses >=12 mg and I2=0% for most GI outcomes. Dose-specific ORs appear ONLY in the abstract (45 mg: nausea OR 11.48 [6.52-20.21], diarrhoea 3.99 [2.07-7.70], discontinuation for GI AEs 10.22 [4.99-20.94]); vomiting is printed with the identical estimate and CI to nausea, an apparent transcription duplication that cannot be checked because the forest plots (Figures S2/S3) are images. Tempering: 'most GI AEs were mild to moderate in severity and did not lead to treatment discontinuation'. Quality moderate not high - 'This study included only five RCTs, which restricts the ability to detect rare AEs' and 'All trials were industry-sponsored and enrolled selected, relatively low-risk participants'; evidence covers one agent (oral orforglipron) generalised to the class by the authors' inference. |
| McKenzie C et al 2025 · Anesth Prog | observational | supports | low | Narrative review for dental anesthesia providers: GLP-1/GIP receptor agonists decrease gastric emptying, increasing satiety but also risk of retained gastric contents, emesis, and aspiration perioperatively. |
| Wharton 2022 · Diabetes Obes Metab | RCT | supports | high | Pooled STEP 1-3 (2117 semaglutide 2.4 mg vs 1262 placebo, 68 wk, double-blind): 'GI AEs were more common with semaglutide 2.4 mg than placebo, with most frequently nausea (43.9% vs. 16.1% of participants), diarrhoea (29.7% vs. 15.9%), vomiting (24.5% vs. 6.3%) and constipation (24.2% vs. 11.1%).' Any GI AE 72.9% vs 47.1%. 'Most GI AEs with semaglutide were non-serious (99.5% of AEs), mild-to-moderate (98.1%), transient and occurred most frequently during/shortly after dose escalation' — median duration nausea 8, diarrhoea 3, vomiting 2 days; constipation 47 days (35 on placebo). Permanent discontinuation for GI AEs 4.3% vs 0.7%. Gastric emptying and gastroparesis were not measured, so this source votes on the outcome, not the claim's mechanism. Weight loss was largely INDEPENDENT of GI AEs (under 1 pp of the 7.6-14.4 pp placebo-corrected loss mediated by them), correcting the prior extract's 'modest extra weight loss in those affected' framing. |
| Manoj RJ et al 2026 · World J Gastrointest Pathophysiol | observational | supports | low | Narrative review, full text: for first-generation GLP-1 RAs 'The primary adverse effects reported were gastrointestinal - nausea, vomiting, diarrhea, and constipation - which often necessitate gradual dose escalations and sometimes led to treatment discontinuation'. Class-wide, 'The most frequently reported adverse events are gastrointestinal symptoms, such as nausea, diarrhea, vomiting, and constipation, which are typically mild-to-moderate and transient'; in phase 3 tirzepatide trials 'up to 81.8% of participants reported at least one adverse event, but most were gastrointestinal', with discontinuation 4.3%-7.1% across doses, and GI events 'remain the most frequent cause of treatment discontinuation - reported in about 10%-17% of patients in oral GLP-1 RA studies and up to 7% for injectable polyagonists'. |
| Kanbay M et al 2026 · Clinical Kidney Journal | observational # was meta-analysis — the GI endpoint was NEVER pooled; authors file it under "Observational cohorts" (~520 pts, 9 single-centre, no comparator) (reread #445) | supports | moderate | Systematic review/meta-analysis in kidney transplant recipients (17 studies, n=54,680 overall) — but the GI signal is NOT pooled and does not rest on that denominator: 'Nine studies reported gastrointestinal (GI) adverse events [ 10–12 , 14–16 , 19 , 22 , 23 ]. Across studies, GI side effects were the most frequently observed adverse events associated with GLP-1 receptor agonists in kidney transplant recipients and were generally mild and self-limited. Reported symptoms included nausea, vomiting, diarrhea, and reduced appetite. The incidence of GI intolerance ranged from ∼10% to 20%.' The nine studies with GI data (Suppl. Table S7) are small single-centre cohorts totalling ~520 patients; the large registry cohorts (Orandi n=18,016; Lin n=3,297; Cohen n=136) reported no GI data. Discontinuation: 'Discontinuation due to GI symptoms was uncommon in most cohorts; however, higher discontinuation rates were observed in isolated studies, primarily driven by persistent intolerance.' and 'Kukla et al . and Vigara et al . reported higher discontinuation rates (29% and 11 of 16 patients, respectively), mainly due to GI symptoms [ 12 , 22 ].' 'No study reported GI-related hospitalizations or severe complications.' Authors' own GRADE row: 'GI adverse events | 9 (~800) | Observational cohorts | ... | Moderate | Mild, self-limited nausea common; few discontinuations.' |
| Sodhi 2023 · JAMA | observational | supports | moderate | Pharmacoepi cohort (weight-loss use): GLP-1RA vs bupropion-naltrexone raised gastroparesis (HR 3.67, 1.15-11.90), bowel obstruction and pancreatitis. |
| Aronne 2025 · N Engl J Med | RCT | supports | moderate | SURMOUNT-5 (NCT05822830, n=751, 72 wk, obesity without diabetes): 'The most common adverse events in both treatment groups were gastrointestinal, and most were mild to moderate in severity and occurred primarily during dose escalation.' Supports the class GI-tolerability claim via the semaglutide arm, but this is an 'open-label' trial with NO placebo arm, so subjective GI symptoms were ascertained unblinded and incidence cannot be attributed to drug versus background; no numeric GI event rates and no gastric-emptying/gastroparesis data available in the abstract. |
| Merhavy ZI et al 2026 · Perioperative Medicine | observational | supports | low | Systematic review of 32 articles (PROSPERO 1,110,268), narrative synthesis only - 'Due to the extensive variability in the type, quality, and consistency of the articles available, a meta-analysis was not able to be conducted.' On the delayed-gastric-emptying endpoint the included literature is consistent: 'Whether shown by a ten-fold, four-fold, five-fold, or 19% increase, studies have consistently reported a higher likelihood of GLP-1 RA users having more residual gastric contents, even after appropriate fasting, compared to non-users' (15-56% of users vs 5-20% of controls), and 'other preprocedural GI effects were observed with a higher incidence in GLP-1 RA users such as reduced gastric acidity, nausea, vomiting, dyspepsia, and abdominal distension (OR 15.1; 95% CI 9.85-23.45).' The downstream aspiration/regurgitation signal is inconsistent, but that is not this claim's outcome. Quality low: 'Many of the studies assessed were of relatively low quality (case studies or small study groups) and were therefore underpowered', the included set mixes narrative reviews and case series with cohorts, and '1,859 could not be accessed through journal paywalls'. |
| Takrori E et al 2025 · Medicina (Kaunas) | observational | supports | moderate | PRISMA/PROSPERO systematic review (12 studies of 733 screened, incl. one cohort of 18,386) of GI adverse effects of anti-obesity drugs in NON-diabetic adults; narrative synthesis only, no pooling. GLP-1RAs carried the highest GI burden: in STEP 1 'semaglutide recipients experienced GI disorders in 74.2% versus 47.9% with placebo' and 'Nausea occurred in 44.2% vs. 17.4%, diarrhea 31.5% vs. 15.9%, vomiting 24.8% vs. 6.6%, and constipation 23.4% vs. 9.5%', with 'Discontinuation due to GI events occurred in 4.5% on semaglutide vs. 0.8% on placebo'. The review attributes this to the claim's stated mechanism — 'Their mechanism of delayed gastric emptying and central appetite suppression explains the predominant symptoms of nausea, diarrhea, vomiting, and constipation' — and finds it dose-dependent and transient: 'most symptoms were mild to moderate and dose-dependent, peaking during escalation and resolving with continuation'. Self-rated certainty for this class is Moderate. Secondary synthesis of trials the vault holds separately (STEP 1/4, SURMOUNT-4/CN, Rodriguez 2024) — not an independent voice. Gastroparesis specifically is not assessed. |
| Wen J, Puglisi J, Frezza E 2025 · Cureus | observational | supports | low | Retrospective single-centre, single-surgeon chart review (Cureus) of 144 elective EGDs: of 12 patients on semaglutide 2.4 mg or tirzepatide 2.5 mg for >6 months, 'Out of the 12 patients, 11 had food particles described in their stomachs during the procedure' (91.7%), while 'Upon review of the 132 patient charts that did not fulfill our patient criteria, there were no descriptions of gastroparesis or RGC' (p<0.001), despite NPO >=12 h. Confirms the claim's delayed-gastric-emptying mechanism as a procedural finding, but weakly: exposed n=12, and the authors state 'Third, there was no control group to determine if the RGC observed was due to GLP-1 RA usage' and that the sample 'prohibits a determination of causal effect despite the statistical significance.' No symptom endpoint (nausea/vomiting/gastroparesis) was measured, and no aspiration occurred: 'All patients underwent EGDs without complications or aspiration and were discharged on the same day.' |
| Lorenz N et al 2025 · Cureus | observational | supports | low | Case report: 52F with no surgical history developed functional small bowel obstruction after tirzepatide (Zepbound) dose escalation, no other identifiable cause. |
| de Mesquita 2023 · Int J Obes | meta-analysis | supports | moderate | Meta-analysis of 6 industry-funded tirzepatide RCTs (1851-2008 participants per endpoint) versus placebo. GI events are raised at every dose: nausea 'OR 3.12 (2.34 to 4.17)' / 'OR 4.59 (3.46 to 6.07)' / 'OR 4.24 (2.40 to 7.48)'; vomiting 'OR 4.04 (2.40 to 6.82)' / 'OR 5.20 (3.12 to 8.67)' / 'OR 6.24 (3.89 to 10.00)' at 5/10/15 mg — a clean dose ladder; diarrhea 'OR 2.04 (1.22 to 3.42)' / 'OR 2.26 (1.35 to 3.78)' / 'OR 2.36 (1.45 to 3.84)'. Constipation and dyspepsia are also higher on drug (direction-only, no pooled estimate given). GRADE certainty runs High (vomiting) to Low (nausea 5 mg, diarrhea 5 and 10 mg), with the table conceding '‡ All included studies were financed by the pharmaceutical industry.' Serious adverse events were 'generally similar between tirzepatide and placebo'. Caveats on the prior extract: no discontinuation data appear anywhere in the retrievable text, and nothing here compares tirzepatide to another GLP-1RA, so neither 'drives most discontinuations' nor 'class-consistent' is supported by this paper. Drug is a dual GIP/GLP-1 receptor agonist, so part of the GI signal may be GIP. |
| Kansakar U et al 2026 · Int J Mol Sci 2026;27(3):1409 | observational | supports | low | Kansakar U, Jankauskas SS, Pande S, Mone P, Varzideh F, Santulli G. Int J Mol Sci 2026;27(3):1409 (MDPI), PMID 41683830, doi 10.3390/ijms27031409. NARRATIVE review of orforglipron - 'No new data were created or analyzed in this study' - so evidence_role: review, zero weight; it restates the Lilly Phase 1-3 / ACHIEVE-1 / ATTAIN programme the vault should count directly. Read from FULL TEXT (JATS, 193 kB) 2026-08-20, upgrading a prior abstract-grade extract. On direction it is unambiguous and repeats the claim in four sections: 'Gastrointestinal adverse events, primarily nausea, vomiting, and diarrhea, were dose-dependent and typically mild to moderate in severity'; 'Gastrointestinal adverse events remain the primary limitation for tolerability'; 'Its observed safety profile has been dominated by dose-dependent gastrointestinal adverse events, consistent with GLP-1R agonism'; and it generalises beyond orforglipron - AEs 'most prominent during the initial titration phase, mirroring the class effects observed with peptide-based GLP-1 RAs such as semaglutide and liraglutide'. The claim's mechanism clause is also affirmed: 'In the gastrointestinal tract, orforglipron slows gastric emptying'. LIMITS that hold quality at low: NOT ONE GI incidence rate appears anywhere in the full text - no percentage for nausea, vomiting or diarrhea, no n, no CI, only bracketed reference numbers and range-valued tables; gastroparesis, part of the claim's object, is never mentioned; and the paper reports a PROSPERO registration (CRD420261293993) while containing no methods, no search strategy and no screening flow. |
| Radparvar I et al 2025 · Archives of Clinical and Medical Case Reports | observational | supports | low | Narrative review (no Methods, no search strategy, no data of its own; 63 references) of incretin-based therapies. On GI adverse events it concludes: 'Gastrointestinal adverse effects are the most common complications for both semaglutide and tirzepatide.' Its numbers are restated from cited primaries — semaglutide vs placebo over 30 weeks in T2D: 'nausea in 11.4 to 20% of the semaglutide-treated patients (placebo 3.3-8%), vomiting in 4 to 11.5% (placebo 2-3%) and diarrhea in 4.5 to 11.3% (placebo 1.5-6%)'; tirzepatide in SURPASS: 'nausea occurred in approximately 12-24% of tirzepatide-treated patients (vs 6-12% with placebo or active comparators), vomiting in 5-14%, and diarrhea in 12-22%'; and pooled across obesity trials 'the likelihood of experiencing any GI side effect was nearly twice as high with either drug. Tirzepatide was associated with a higher risk than semaglutide, with a RR of 2.94 compared with 1.68 for semaglutide.' Direction is unambiguous, but the evidence is second-hand and descends from the same trial programmes the vault cites elsewhere, so this counts as a review, not an independent primary vote. |
| Rallabandi S et al 2026 · Cureus | observational # was n-of-1 (weight 5) — uncontrolled case report, not a structured n-of-1 trial; same fix as reread #434 | supports | low | Case report (n=1, Cureus): 59F, BMI 43, on semaglutide for weight loss; 3 days after escalation 0.25->0.5 mg weekly she developed nausea, vomiting and abdominal pain, and laparoscopy 'showed pan dilatation of small and large bowels without adhesions, hernia, or masses, suggesting severe ileus' (functional SBO), with AKI from hypotensive acute tubular necrosis (creatinine 1.58->3.44 mg/dL in 48 h) requiring weeks of hemodialysis. Causality is weak by the authors' own instrument: Naranjo score 4 = 'possible' only, heavy confounding (granulomatosis with polyangiitis on azathioprine/rituximab, diverticulosis, BP 77/55), and they state the patient's death from recurrent ileus four months after semaglutide discontinuation 'challenges a causal interpretation'. |
| Crisafulli S et al] # corrected 2026-08-17 from PubMed; was [2026 2026 · Ann Intern Med | observational | mixed | high | OFF-SCOPE for a causation claim. New-user, active-comparator claims cohort study with three PAIRWISE INCRETIN-VS-INCRETIN comparisons and no non-incretin (or non-user) arm: 'New-user, active-comparator cohort study' in 'Adults with T2D initiating dulaglutide, subcutaneous semaglutide, and tirzepatide between 1 January 2019 and 30 August 2024 in 3 cohorts corresponding to 3 pairwise comparisons.' Very large and well matched - 'There were 65 238 matched pairs in the semaglutide versus dulaglutide cohort, 20 893 in the tirzepatide versus dulaglutide cohort, and 46 620 in the tirzepatide versus semaglutide cohort', 1:1 propensity-score matched - and the result is a flat set of between-drug hazard ratios: 'The HR of gastrointestinal events was 0.96 (95% CI, 0.87 to 1.06) in the semaglutide versus dulaglutide cohort, 0.96 (CI, 0.77 to 1.20) in the tirzepatide versus dulaglutide cohort, and 1.07 (CI, 0.90 to 1.26) in the tirzepatide versus semaglutide cohort.' Two reasons this cannot vote on 'GLP-1RAs cause GI side effects'. (1) The denominator is another GLP-1RA, so an HR near 1.0 is silent on whether either drug raises GI risk above no drug; a design with no unexposed comparator cannot estimate the drug-attributable effect the claim asserts. (2) The measured outcome is a SEVERE-event composite - 'The primary outcome was a composite of acute pancreatitis, biliary disease, bowel obstruction, gastroparesis, and severe constipation' - not the nausea and vomiting the claim names; only the 'rare gastroparesis' clause overlaps at all, and even that is reported only as a between-drug ratio. Absolute incidence rates, which could have spoken to the rarity clause, are not in the abstract and full text could not be retrieved - cannot-tell from available text. Methodologically strong for its own question (Ann Intern Med; NIDDK-funded, not industry; limitation conceded as 'Possible residual confounding by glycemic control and body mass index'), which is why quality stays high; the role, not the quality, is what was wrong. Its real finding is comparative parity: 'These findings suggest that dulaglutide, semaglutide, and tirzepatide have similar gastrointestinal safety profiles in adults with T2D.' |
| Nahar S, Maybee N, Tamanna N, Sadat A, Khanam F, Begum R, Akther S, Khan MS, Sonia SN, Hasan N 2025 · Cureus 2025;17(12):e98935 | observational # was n-of-1 (weight 5) — an uncontrolled case report is not a structured n-of-1 trial; matches source study_type (reread #434) | supports | low | Single uncontrolled case report (Cureus), 61F with T2DM, cryptogenic cirrhosis, CKD-3 and prior post-hysterectomy SBO, on tirzepatide 5 mg weekly for >1 year. On-drug GI motility signal is real: 'She reported chronic constipation, with bowel movements only once or twice weekly for several months before admission, a change from her baseline of daily to every-other-day bowel movements.' She then had a closed-loop obstruction with 25 cm of necrotic bowel resected. But the causal attribution is weaker than a supporting vote implies, and the paper says so: 'Although adhesive disease was the direct cause of the obstruction, the temporal relationship... suggests a contributory role', 'the Naranjo scale yielded a score of 4, indicating a possible adverse drug reaction', and the conclusion states the association 'suggests a possible contributory role of GLP-1 RA-induced motility changes, rather than establishing direct causality.' Six constipating drugs were on board (amitriptyline, gabapentin, duloxetine, citalopram, quetiapine, tramadol), which the authors say 'likely compounded her baseline risk for constipation and impaired GI motility.' Note also that the claim's named outcomes were absent here: 'She denied any vomiting, hematemesis, melena, or hematochezia.' Onset was ~15 months after initiation; no rechallenge, and the event was resolved surgically, not by dechallenge. |
Disagree, or know a study we missed?
We grade by evidence, not opinions. The way to weigh in is to point us to a study we haven't cited (check the evidence table above first), or to flag a problem with one we have. Every submission is reviewed; if it holds up, the grade updates and shows in Science Changes Its Mind.
Opens a short form. You'll sign in with Google so submissions are tied to a real account — we don't display your identity, and we only accept a link we can verify (PubMed, DOI, ClinicalTrials.gov).
Educational only, not medical advice. Grades and scores reflect published evidence weighted by study design and quality; see the methodology.