Residual gastric content and aspiration-related outcomes in perioperative patients treated with semaglutide: A systematic review and meta-analysis

Abstract

BACKGROUND

Semaglutide, a long-acting glucagon-like peptide-1 receptor agonist, is increasingly prescribed for type 2 diabetes and obesity. Its perioperative safety remains uncertain due to concerns regarding delayed gastric emptying, increased residual gastric content (RGC), and possible aspiration-related complications.

AIM

To determine whether perioperative semaglutide use is associated with RGC, pulmonary aspiration, digestive symptoms, procedure discontinuation, and gastroparesis versus controls.

METHODS

We conducted a systematic review and meta-analysis in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses and Cochrane guidelines. Seven electronic databases (Scopus, Web of Science, EMBASE, MEDLINE/PubMed, Latin American and Caribbean Health Sciences Literature, Cochrane Library, and OpenGrey) were searched from inception through April 2026. The methodological quality of observational studies was assessed using the Newcastle-Ottawa Scale, and the certainty of evidence for each outcome was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation approach.

RESULTS

Twelve observational studies (49651 patients) were included; no randomized trials were identified. Methodological quality was generally high (Newcastle-Ottawa Scale 8-9), although residual confounding and selection bias remain inherent limitations of observational evidence. Semaglutide use was associated with increased risks of RGC [relative risk (RR) = 3.49, 95% confidence interval (CI): 1.78-6.83; moderate certainty)] and digestive symptoms (RR = 5.65, 95%CI: 3.61-8.84; moderate certainty). For pulmonary aspiration (RR = 1.46, 95%CI: 0.29-7.24; very low certainty), procedure discontinuation (RR = 1.74, 95%CI: 0.59-5.11; low certainty), and gastroparesis (RR = 1.43, 95%CI: 0.71-2.87; low certainty), the pooled estimates were imprecise and not statistically significant, and clinically important differences could not be excluded. Heterogeneity regarding RGC was substantial (I² = 87.8%), which limits the interpretability of the pooled estimate.

CONCLUSION

Current observational evidence suggests that perioperative semaglutide use increases RGC and digestive symptoms, but clinical significance remains uncertain; aspiration-related outcomes require cautious interpretation and prospective standardized studies.

Key Words: Semaglutide; Glucagon-like peptide-1 receptor agonists; Residual gastric content; Pulmonary aspiration; Perioperative management; Gastric ultrasound

Core Tip: Perioperative semaglutide treatment was associated with an increased risk of residual gastric content and digestive symptoms. Evidence regarding pulmonary aspiration, gastroparesis, and procedure discontinuation remains uncertain because the available data are observational and imprecise. Prospective studies using standardized fasting, perioperative, and outcome definitions are needed.

INTRODUCTION

Glucagon-like peptide (GLP)-1 receptor agonists (GLP-1RAs) have become central in treatment for obesity and type 2 diabetes[1,2]. According to the World Obesity Atlas 2025, more than 1 billion people worldwide are living with obesity[3]. The 11th edition of the International Diabetes Federation’s Diabetes Atlas (2025) estimates that 589 million adults aged 20-79 years were living with diabetes in 2024, and more than 90% of these cases were type 2 diabetes[4].

Semaglutide, a long-acting GLP-1RA, is now widely used in both metabolic and obesity care. In addition to enhancing glucose-dependent insulin secretion and suppressing glucagon release, GLP-1 receptor agonism delays gastric emptying[5,6]. Experimental and clinical data indicates that this effect is mediated by relaxation of the gastric fundus, increased gastric compliance, inhibition of antral contractility, and increased pyloric tone[7]. Although this effect may be attenuated over time due to tachyphylaxis, delayed gastric emptying may persist in some patients and has raised concern about residual gastric content (RGC), regurgitation, and pulmonary aspiration during anesthesia or procedural sedation[8-10].

These potential complications could increase perioperative morbidity and prolong recovery. Current recommendations for the perioperative management of GLP-1RAs are based on limited evidence, primarily from case reports and small observational studies[6,11,12]. Therefore, this systematic review and meta-analysis was conducted to determine whether preoperative risk among adult patients undergoing surgery or procedural sedation is higher with or without semaglutide use. The primary outcome was RGC. Secondary outcomes included digestive symptoms, pulmonary aspiration, procedure discontinuation, and gastroparesis.

MATERIALS AND METHODS

Protocol and registration

This study was conducted according to Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines and the Cochrane handbook[13,14]. The systematic review protocol was registered with the International Prospective Register of Systematic Reviews, maintained by York University, approval No. CRD42024609601.

Search strategy and study selection

A search from inception to April 2026 was conducted in seven electronic databases: Scopus, Web of Science, EMBASE, MEDLINE (PubMed), Latin American and Caribbean Health Sciences Literature, Cochrane Library for Systematic Reviews, and Opengray.eu. Studies retrieved from these databases were systematically and independently reviewed by two researchers (Vinicius Remus Ballotin and Pedro Lucas Carneiro Ferreira), and the reference lists of the retrieved studies were submitted for manual search. Divergences in study selection were resolved by a third researcher (Carolina da Silva Borges). The full search strategy is described in the Supplementary material.

Eligibility criteria

The inclusion criteria were as follows: Adults aged ≥ 18 years; patients who received preoperative semaglutide treatment, regardless of therapy duration, before surgery; and randomized controlled trials and observational studies (including cohort, case-control, and cross-sectional designs) reporting at least one of the following perioperative outcomes: (1) RGC (primary outcome); (2) Digestive symptoms; (3) Pulmonary aspiration; (4) Procedure discontinuation; and (5) Gastroparesis.

The exclusion criteria were as follows: Significant comorbidities that independently increase the risk of perioperative complications (e.g., severe cardiovascular or respiratory diseases); case reports, case series, reviews, and expert opinions that do not present original data or comparisons between groups; non-English language studies without readily available translated versions; and studies with significant missing data that impede appropriate analysis and interpretation of outcomes.

Methodological quality assessment

Study quality was assessed using the Newcastle-Ottawa Scale (NOS)[15,16], depending on type. Two researchers (Remus Ballotin V and Carneiro Ferreira PL) conducted independent assessments, and their results were compared to evaluate interrater reliability. Divergences were discussed with a third researcher (Tonin de Almeid V) until 100% consensus was reached. The NOS is an assessment tool that examines the quality of observational studies across selection, comparability, and outcomes, allowing for the classification of risk of bias and internal validity.

Certainty of evidence assessment

The certainty of evidence for each outcome was evaluated using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach[17,18]. For each outcome, the overall certainty of evidence was rated as high, moderate, low, or very low, according to GRADE guidelines.

Data extraction

Two researchers (Tonin de Almeida H, Giovanardi Pandolfo da Silva R) independently extracted qualitative and quantitative data using a predefined form. Any discrepancies were reviewed by a third researcher (Remus Ballotin V). The extracted qualitative data included the: (1) Study aim; (2) Databases searched; (3) Population, interventions, and outcomes; and (4) Authors’ conclusions. The quantitative data extracted comprised: (1) The number of patients; (2) Population characteristics (e.g., number of participants, percentage of females, and mean age); and (3) Effect-size results for perioperative outcomes comparing the semaglutide group with the control group. When an eligible study enrolled a broader GLP-1RA population, only data from the semaglutide-specific subgroup were extracted for this review. If semaglutide-specific population characteristics or outcome data were not separately reported, the corresponding authors were contacted to request disaggregated data. Studies from which semaglutide-specific data could not be isolated were excluded from the quantitative synthesis. This approach was taken to preserve exposure specificity and improve comparability across studies by ensuring that pooled estimates reflected semaglutide-specific rather than mixed GLP-1RA effects. When overlapping populations were suspected, only the most complete or most recent report was retained.

Statistical analysis

Data were synthesized through a meta-analysis of studies reporting perioperative outcomes associated with preoperative semaglutide use. All outcomes were extracted as the number of events and total participants in each group, allowing effect estimates to be standardized as risk ratios (RR). For pulmonary aspiration, the sample size required to detect a small absolute difference in risk was additionally estimated using a standard power calculation in G*Power 3.1.9.7. When at least two studies reported comparable outcomes, pooled estimates were calculated using both fixed-effect and random-effects models. Given the expected clinical and methodological heterogeneity across studies, the random-effects model was considered the primary analysis, whereas fixed-effect estimates were used as complementary analyses. Fixed-effects analyses employed the Mantel-Haenszel method. Random-effects meta-analyses used an inverse-variance approach with the Paule–Mandel estimator for between-study variance (τ²) and the Q-profile method for τ² confidence intervals (CI). To provide more conservative estimates under heterogeneity, 95%CIs for pooled effects were calculated with the Knapp-Hartung adjustment. Statistical heterogeneity was assessed using the I² statistic, and subgroup analyses were performed according to study design. Publication bias was evaluated visually using funnel plots and statistically with Egger’s test. The analyses were conducted in R Studio 2024.12.1, with forest plots generated for visual representation. In sparse-event analyses, zero cell frequencies were calculated by the software using a continuity correction of 0.5, being applied only to calculate individual study estimates. In the sensitivity analyses, studies at higher risk of bias were excluded, and leave-one-out influence analyses were used.

Data and resource availability

The data underlying this study are available in the Figshare repository, accessible at https://doi.org/10.6084/m9.figshare.30450836[19].

RESULTS

Study selection

The database search yielded 16159 records. After removing duplicates, 13377 titles and abstracts were screened, of which 13259 were excluded. A total of 112 full-text articles were assessed for eligibility, of which 100 were excluded, leaving 12 studies in the final analysis. The reasons for exclusion are summarized in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis flowchart (Figure 1) and detailed in the Supplementary material.

Figure 1  Preferred reporting items for systematic reviews and meta-analysis flowchart.

Quality assessment

NOS scores ranged from 8 to 9 (maximum of 9), with an initial interrater agreement of 92.3%. Most achieved maximum scores in the selection and outcome domains, reflecting representative cohorts, reliable exposure measurement, and valid outcome assessment. Minor reductions in total score were generally caused by inadequate follow-up or comparability. Given their observational design, these studies remain at inherent risk of residual confounding and selection bias, despite generally high NOS scores.

Characteristics of the included studies

Overall, 12 studies were included, comprising ten cohort studies, of which one was a prospective study[11] and two were case-control studies[20,21]. Most studies were published in 2024, except for one published before and two published after 2024[10,21,22]. Most of the studies were conducted in the United States, followed by Brazil, Canada, and Belgium. The largest study included 23940 participants[23], whereas the smallest included 88[22]. The percentage of female participants could not be assessed in three studies[11,20,23], and in one study, data were available only for the control group[11]. The mean age ranged from 45 years to 71 years in the intervention groups and from 39 years to 72 years in the control groups. The reported procedures included upper digestive endoscopy, colonoscopy, and elective surgeries, such as joint prosthesis implantation, cardiovascular surgery, and digestive tract surgery. Many studies did not report the semaglutide holding interval[8,21,23-25]. One study classified RGC content as small, medium, or large[26]. Two studies did not report RGC data[12,23]. Apart from RGC, the most commonly reported outcomes were pulmonary aspiration, digestive symptoms, procedure discontinuation, and gastroparesis. Only one study reported on 30-day mortality, postoperative length of stay, and pneumonia[12]. Further details are provided in Tables 1 and 2.

Table 1 Baseline characteristics of patients in the included studies (reported data only) (total n = 49651), n (%)/mean ± SD.

Variable	Semaglutide	Control
Patients	11022 (22.2)	38629 (77.8)
Sex (female)	2514 (45.5)	8785 (43.9)
Age (mean years)	56.7 ± 7.41	56 ± 8.95
BMI (mean)	31.58 ± 4.54	29.2 ± 3.28
ASA
I	69 (0.6)	313 (0.8)
II	214 (1.94)	1094 (2.8)
III	8 (0.07)	184 (0.4)
IV	0 (0)	1 (0.002)
V	0 (0)	0 (0)
Comorbidities
Type 2 diabetes	4631 (42.3)	15253 (40.8)
GERD	917 (8.3)	3313 (8.8)
Opioid use	1501 (13.7)	5712 (15.2)
Abdominal surgery	3	51
Cirrhosis	15	11
HbA1c (mean)	6.48 ± 0.58	6.42 ± 0.67
	47 mmol/mol	47 mmol/mol
Duration (mean day)	149 ± 127	-
Dose (mean mg)	1.33 ± 0.23	-
Fasting (mean hours)
Solids	12.89 ± 3.28	12.83 ± 3.27
Clear fluids	6.51 ± 3.14	7.66 ± 3.8

Values are based on available reported data. Not all studies reported every baseline variable; thus, denominators may differ between characteristics and from those used in the pooled effect-size calculations for specific outcomes (e.g., residual gastric content and aspiration). BMI: Body mass index; ASA: American Society of Anesthesiologists Physical Status classification; GERD: Gastroesophageal reflux disease; HbA1c: Glycated hemoglobin.

Table 2 Characteristics of the included studies.

Ref.	Country	Study type	Inclusion criteria	Exclusion criteria	Fasting	Procedure indication	Semaglutide interruption	Increased RGC definition	Anesthesia details	Extracted data source	NOS score
Santos et al[26], 2024	Brazil	Retrospective cohort	≥ 18 years-old, presenting for elective esophagogastroduodenoscopy	Gastric volvulus/frank outlet obstruction/active esophageal/gastric/duodenal bleeding, ASA physical status ≥ IV, recent (≤ 2 months) abdominal surgery, emergency procedures, EGD combined with surgical procedures, chronic renal/Liver disease, achalasia, Zenker’s diverticulum, linitis plastica, multiple myeloma, systemic collagenosis, amyloidosis, pregnancy, chronic opioid use, drug addiction, use of vasoactive agents, intensive care patients, preoperative use of medication that affect gastric emptying (tricyclic antidepressants, opioids, pro-kinetics, histamine H2- receptor antagonists) other than semaglutide, and incomplete medical records. Patients using GLP-1-Ras other than semaglutide and/or oral/daily semaglutide were also excluded	≥ 2 hours for clear fluids, ≥ 8 hours for solids and fluids with residue	General indications for EGD included dysphagia, odynophagia, persistent abdominal pain, intractable and/or chronic gastro-esophageal reflux	Preoperative interruption intervals were categorized based on its ~7-day half-life as ≤ 7 days, 8-14 days, 15-21 days, and > 21 days	Any amount of solid content from the esophagus to the pylorus, or > 0.8 mL/kg of fluid content (given its higher risk of bronchoaspiration) as measured from the aspiration/suction canister	Premedication was not routinely administered. The sedation/anesthetic procedure was at the discretion of the attending anesthesiologist and generally consisted of propofol titration to maintain spontaneous ventilation (under supplemental oxygen via nasal cannula), with occasional tracheal intubation. Intraoperative monitoring included sphygmomanometer, electrocardiography, pulse, and capnography	Main group	High
Phan et al[24], 2024	United States	Retrospective cohort	Patients on confirmed GLP1-Ras receiving an EGD between 2021 and 2023	NA	NA	NA	NA	NA	NA	Subgroup	High
Nersessian et al[2], 2024	Brazil	Retrospective cohort	All patients aged ≥ 18 years presenting for elective surgery between July and December 2023 at the Sao Luiz Hospital/Rede D’Or, Sao Paulo, Brazil	Type 2 diabetes; hiatal hernia; previous gastric surgery (gastrectomy, Roux-en-Y gastrojejunostomy, gastric band, fundoplication); ASA physical status ≥ 3; BMI > 40 kg/m2; recent (within 2 months) abdominal surgery; pre-operative semaglutide interruption interval > 10 days; chronic renal and/or liver disease; achalasia; Zenker’s diverticulum; linitis plastica; multiple myeloma; systemic lupus erythematosus and other collagenosis; pregnancy; chronic opioid use; drug addiction; pre-operative use of medication known to affect gastric emptying other than semaglutide (e.g., tricyclic antidepressants and opioids); and GLP-1 receptor agonists other than semaglutide	≥ 2 hours for clear fluids and ≥ 8 hours for solids and fluids with residue	NA	Patients were grouped according to time since their last dose of semaglutide, either within 1-7 days or 8-10 days	Any amount of solid content or > 1.5 mL/kg of clear fluids	NA	Main group	High
Gu et al[20], 2024	United States	Case-control	Adult patients on semaglutide who underwent an EGD between August 2022 and August 2023	Patients were excluded if they were in the intensive care unit at time of EGD, previously on GLP-1 agonist therapy, previously diagnosed with gastroparesis, or underwent EGD for acute gastrointestinal bleeding	NA	NA	NA	NA	NA	Main group	High
Zaffar et al[8], 2024	United States	Retrospective cohort	Of 2578 EGDs performed on adults (age 18-89 years) under deep sedation/general anesthesia between August 2022 and August 2023 were included in the study after simple random sampling	NA	NA	NA	NA	RGC found during EGD	Patients under deep sedation/general anesthesia	Subgroup	High
Silveira et al[10], 2023	Brazil	Retrospective cohort	All patients ≥ 18 years-old presenting for elective diagnostic upper endoscopy were eligible	Exclusion criteria were: Gastric outlet obstruction, gastric volvulus, frank/active esophageal/gastric/duodenal bleeding, ASA physical status ≥ IV, recent (≤ 2 months) abdominal surgery, emergency endoscopic procedures, urethral erosion combined with other/surgical procedures, chronic renal and/or liver disease, achalasia, Zenker’s diverticulum, linitis plastica, multiple myeloma, systemic collagenosis, amyloidosis, pregnancy, chronic opioid use, drug addiction, use of vasoactive agents, patients admitted to the intensive care unit, preoperative use/ingestion of medication known to affect gastric emptying (e.g., tricyclic antidepressants, opioids, pro-kinetics, histamine H2-receptor antagonists) other than semaglutide, and incomplete medical records. Patients using GLP-1 agonists other than semaglutide were also excluded	≥ 2 hours for clear fluids, and ≥ 8 hours for solids and fluids with residue	Elective diagnostic upper endoscopy	The time intervals of semaglutide interruption in patients with and without increased residual gastric content were 10 (6-15) and 11 (7.75-12.5) days, respectively (P = 0.67)	Any amount of solid content from the esophagus to the pylorus, or > 0.8 mL/kg of fluid content as measured from the aspiration/suction canister	The sedation/anesthetic procedure was at the discretion of the anesthesiologist	Main group	High
Korlipara et al[25], 2024	United States	Retrospective cohort	Adults undergoing EGD at Weill Cornell from January 2018 to March 2023, with or without semaglutide	NA	NA	NA	NA	Defined as “retained gastric contents” – no specific numeric threshold given	NA	Main group (author-provided)	High
Alkabbani et al[23], 2024	United States	Retrospective cohort	Patients using GLP-1 receptor agonists or SGLT-2 inhibitors undergoing upper endoscopy	Not clearly listed; based on claims data filters and diagnostic codes	NA	NA	NA	Implicitly assessed via diagnostic codes for aspiration and procedure discontinuation	NA	Subgroup	High
Welk et al[12], 2024	Canada	Retrospective cohort	Patients ≥ 66 years with type 2 diabetes who underwent elective surgery under general or spinal anesthesia between February 2020 and March 2023	Emergency surgery, multiple surgeries on the same day, incomplete or inconsistent data	Exclusion criteria included emergency surgery, multiple surgeries on the same day, and incomplete or inconsistent data	Various elective surgeries (e.g., joint replacement, cardiovascular, digestive, etc.)	None – patients were on active semaglutide therapy at the time of surgery	Not directly assessed – the outcome was postoperative pneumonia as an indirect marker of aspiration (RGC was neither visualized nor measured)	General anesthesia (58.6%) or spinal anesthesia (41.2%); data on type of induction, airway management, and medications were not included in the involved database	Main group	High
Ukwade et al[21], 2025	United States	Retrospective	EGD procedures performed between September 1, 2022, and October 30, 2023	Patients who were pregnant, incarcerated, under the age of 18, patients with a BMI > 50, and patients who were hospitalized. Patients were also excluded if they met the ASA physical status classification of ASA 4 (patients with severe systemic disease threatening life) or ASA 5	Midnight fast before EGD; morning medications allowed with small sips of water. Those who consumed any food are rescheduled as, per our institutional guidelines, since a fast of 8 hours is XXXrequire for any food other than clear liquids and a fast of 2 hours is XXXrequire for any clear liquids prior to an EGD	NA	NA	Defined as mentioning food or fluid in the endoscopy report, which was based on the endoscopist’s clinical judgment	NA	Subgroup	High
Vlaeminck et al[22], 2026	Belgium	Prospective cohort	Adult patients receiving semaglutide treatment – regardless of dose, administration route, frequency, or therapeutic indication - who were scheduled for elective surgery under general anesthesia	Patients were not included if they declined to participate; had a contraindication to gastric ultrasound (i.e., previous gastric surgery or hiatal hernia); comorbidities known to delay gastric emptying (scleroderma; systemic lupus erythematosus; hypothyroidism; Parkinson’s disease; cerebral palsy; and multiple sclerosis; or were unable to assume the right lateral decubitus position for sonography	European Society of Anaesthesia and Intensive Care fasting guidelines (i.e., > 2 hours for liquids and > 6 hours for solid foods)	NA	The ASA recommended GLP-1 RA withholding period (i.e., 1 week if administered weekly and 1 day if administered daily)	A patient was considered to have a “full stomach” or a “positive” gastric ultrasound if solid gastric content was visible in any position or if the calculated gastric volume in the right lateral decubitus position exceeded 1.5 mL/kg of total body weight	NA	Main group	High

ASA: American Society of Anesthesiologists; BMI: Body mass index; EGD: Esophagogastroduodenoscopy; GLP-1: Glucagon-like peptide 1; NOS: Newcastle-Ottawa Scale; NA: Not available; RGC: Residual gastric content; SGLT-2: Sodium glucose co-transporter 2.

Population

The final analysis included 49651 patients, of whom 11022 (22.2%) were reported as semaglutide users. No overlapping patient populations were identified among the included studies. The mean age in the semaglutide group was 56.7 (SD ± 7.41) years, which was similar to that of the control group at 56 (SD ± 8.95) years. The semaglutide and control groups were 45.5% and 43.9% female, respectively. The mean body mass index was 31.58 (SD ± 4.54) for the semaglutide group and 29.2 (SD ± 3.28) for the control groups. Type 2 diabetes was reported in 4631 (42.3%) patients in the intervention group and 15253 patients (40.8%) in the control group. Most patients in both groups were classified as having American Society of Anesthesiologists physical status I or II; however, several studies did not report American Society of Anesthesiologists classification data. The mean duration of semaglutide use was 149 (SD ± 127) days. The perioperative drug holding interval was reported for most patients, being < 21 days in most cases when data were available.

Meta-analysis

RGC was reported in 239 of 1690 semaglutide users (14.1%) and 322 of 6106 controls (5.2%), corresponding to an absolute risk difference of 8.9 percentage points and a pooled RR of 3.49 (95%CI: 1.78-6.83). There was considerable heterogeneity among the studies (I² = 87.8%, P < 0.0001), with τ² = 0.72 (95%CI: 0.26-2.83), indicating substantial variability in effect sizes. The forest plot in Figure 2A summarizes the results[2,8,10-12,20-26].

Figure 2 Forest plot for residual gastric content and pulmonary aspiration in relation to semaglutide use. A: Residual gastric content; B: Pulmonary aspiration. RR: Relative risk; CI: Confidence interval.

Egger’s test for funnel plot asymmetry found no significant evidence of publication bias (t = 1.33, degrees of freedom = 8, P = 0.21). In the regression model used for this test, residual heterogeneity remained substantial (τ² = 7.51), which suggests that the true effects differed between studies beyond sampling error. Influence analysis using a leave-one-out approach revealed that the overall effect estimate remained robust, with pooled RRs ranging from 2.48 to 4.56 after sequential exclusion of individual studies; all differences remained statistically significant (P < 0.0001). This indicates that no single study disproportionately influenced the overall findings. Details in Supplementary Figure 1.

Digestive symptoms occurred in 27 of 156 semaglutide users (17.3%) and 43 of 1342 controls (3.2%), corresponding to an absolute risk difference of 14.1 percentage points and a pooled RR of 5.65 (95%CI: 3.61-8.84). Conversely, pulmonary aspiration was rare in both groups, occurring in 15 of 6222 semaglutide users (0.24%) and 80 of 21519 controls (0.37%), yielding a highly imprecise pooled RR of 1.46 (95%CI: 0.29-7.24). The forest plot in Figure 2B summarizes these results[2,8,10,23,24]. Details in Supplementary Figure 2.

Procedures were discontinued in 84 of 5979 semaglutide users (1.4%) and in 100 of 18776 controls (0.5%), with an RR of 1.74 (95%CI: 0.59-5.11). Gastroparesis was reported in 20 of 663 semaglutide users (3.0%) and 81 of 2824 controls (2.9%), with an RR of 1.43 (95%CI: 0.71-2.87). These estimates are based on relatively few events and remain highly uncertain, with wide CIs compatible with both no effect and clinically important differences. Secondary outcomes were not often reported. Detailed event counts and study-level estimates are provided in the Supplementary Figures 3 and 4.

Subgroup analysis

Given the high heterogeneity, subgroup analyses were conducted to explore the effect of semaglutide on RGC by study design. The pooled RR across all included studies was significantly high (random-effects model RR = 3.49, 95%CI: 1.78-6.83; P = 0.0008), with considerable heterogeneity (I² = 87.8%).

When stratified by study design, retrospective cohort studies (n = 7) showed a significant association (RR = 3.45, 95%CI: 1.28-9.35), although with high heterogeneity (I² = 91.5%). The single prospective cohort study reported a similar effect size (RR = 2.65, 95%CI: 1.46-4.79). The two case-control studies showed a larger effect estimate in the common-effect model (RR = 4.06, 95%CI: 2.15-7.66). Statistical tests for subgroup differences indicated a lack of significant variation between study designs (random-effects model P = 0.73), suggesting that the association between semaglutide use and increased RGC is consistent regardless of study type (Supplementary Figure 5).

Subgroup analysis by procedure type demonstrated that semaglutide use was associated with an increased risk of RGC across all categories. For endoscopy, the pooled random-effects RR was 2.71 (95%CI: 0.63-11.64; I² = 92.9%), which did not reach statistical significance. In elective procedures, semaglutide users had a significantly higher risk (RR = 4.36, 95%CI: 0.4-46.98; I² = 74%). Conversely, one study assessing combined endoscopy and colonoscopy found a significantly reduced risk [odds ratio (OR) = 0.41, 95%CI: 0.23-0.73][25]. The overall pooled analysis across procedure types yielded an RR of 3.49 (95%CI: 1.78-6.83; I² = 87.8%). Although the test for subgroup differences was significant in the fixed-effect model (P < 0.001), it was not significant in the random-effects model (P = 0.54), indicating that between-study variability might explain the observed differences between procedure types (Supplementary Figure 6).

In a very exploratory meta-regression, the pulmonary aspiration rate did not significantly modify the association between semaglutide use and RGC (P = 0.82). The analysis included only four studies and was underpowered, with unstable coefficient estimates with wide CIs.

Sensitivity analysis

A leave-one-out sensitivity analysis was conducted to evaluate the robustness of the meta-analysis findings on RGC associated with semaglutide use. Sequential exclusion of each study showed that the overall pooled RR remained significant across all analyses, with common-effect model RRs ranging from 2.48 to 4.56 (all P < 0.0001). Likewise, random-effects model estimates remained significant, ranging from 3.03 to 4.25.

Heterogeneity remained substantial throughout the sensitivity analyses, with I² values consistently above 63.2%, mostly exceeding 89.1%, indicating persistent between-study variability not explained by the exclusion of any single study. The τ² estimates ranged from 0.35 to 0.82, confirming this residual heterogeneity.

These results revealed that no single study disproportionately influenced the overall effect estimate, confirming the strength of the association between semaglutide use and increased RGC.

GRADE assessment

Certainty of evidence, assessed using the GRADE approach, ranged from moderate to very low across outcomes. Semaglutide use was associated with a higher RGC risk (RR = 3.49, 95%CI: 1.78-6.83; moderate certainty) despite substantial heterogeneity. Digestive symptoms were also more frequent among semaglutide users (RR = 5.65, 95%CI: 3.61-8.84; moderate certainty), supported by a large, consistent effect size. Evidence for gastroparesis was limited (RR = 1.43, 95%CI: 0.71-2.87; low certainty), as CIs spanned potential benefit and harm. Pulmonary aspiration remained highly uncertain due to very few events and wide CIs (RR = 1.46, 95%CI: 0.29-7.24; very low certainty). Similarly, procedure discontinuation showed an imprecise association (RR = 1.74, 95%CI: 0.59-5.11; low certainty). Details are provided in Table 3.

Table 3 Summary of findings - semaglutide and perioperative outcomes.

Outcome	Participants (total, studies)	Events (SEMA)	Events (control)	Relative effect (RR, 95%CI)	Certainty of evidence (GRADE)
Residual gastric content	7796 (10 studies)	239	322	3.49 (1.78-6.83)	●●◯◯ moderate1
Digestive symptoms	1498 (2 studies)	27	43	5.65 (3.61-8.84)	●●◯◯ moderate2
Gastroparesis	3487 (2 studies)	20	81	1.43 (0.71-2.87)	●◯◯◯ low3
Pulmonary aspiration	27741 (6 studies)	15	80	1.46 (0.29-7.24)	◯◯◯◯ very low4
Procedure discontinuation	24755 (2 studies)	84	100	1.74 (0.59-5.11)	●◯◯◯ low5

¹Residual gastric content: Downgraded one level for serious inconsistency (I² = 90%, high heterogeneity).

²Digestive symptoms: Observational evidence, upgraded one level for large and consistent effect size; final rating = moderate.

³Gastroparesis: Downgraded for serious imprecision [wide confidence interval (CI), few events].

⁴Pulmonary aspiration: Downgraded two levels for inconsistency (variation in effect direction) and very serious imprecision (extremely wide CI, rare outcome).

⁵Procedure discontinuation: Downgraded for inconsistency and imprecision (heterogeneous effects, wide CI).

SEMA: Semaglutida; RR: Relative risk; CI: Confidence interval; GRADE: Grading of Recommendations Assessment.

DISCUSSION

In this systematic review and meta-analysis of observational studies, preoperative semaglutide use was associated with increased risks of RGC and digestive symptoms. In contrast, the associations with pulmonary aspiration, procedure discontinuation, and gastroparesis remained uncertain because these outcomes were infrequently reported, with few events and wide CIs. Overall, the available evidence suggests that semaglutide is associated with intermediate perioperative markers of delayed gastric emptying, whereas evidence for rarer but clinically more consequential outcomes remains limited.

In these 12 observational studies, preoperative semaglutide use was associated with a 3.5-fold higher risk of increased RGC (RR = 3.49, 95%CI: 1.78-6.83) and a 5.6-fold higher risk of digestive symptoms (RR = 5.65, 95%CI: 3.61-8.84), with moderate-certainty evidence. However, substantial heterogeneity limits the interpretability of the pooled effect size estimates. Conversely, the associations with gastroparesis, procedure discontinuation, and pulmonary aspiration were highly uncertain due to few events and wide CIs, rated low to very low certainty.

Most semaglutide users in the included cohorts were overweight or obese middle-aged patients who were undergoing elective endoscopy or surgery, often for weight loss or metabolic indications[12,24,26]. As such, our findings are most applicable to this growing population of relatively stable outpatients.

The absolute risk increase of 8.9 percentage points for RGC suggests that approximately one additional patient in every 11 perioperative semaglutide users will have RGC compared with nonusers. Similarly, the 14.1 percentage point increase in digestive symptoms corresponds to roughly one additional symptomatic patient for every 7 semaglutide users.

Evidence suggests that tachyphylaxis may occur with prolonged therapy, particularly regarding its effect on gastric emptying[2,10,23]. In the included studies, the mean administered dose was 1.33 mg and the mean treatment duration was 149 days. This duration indicates that most patients were transitioning between the titration and maintenance phases of therapy. However, perioperative patients may present at different phases of dose escalation[26,27], which could influence gastrointestinal tolerance and the risk of RGC.

Although semaglutide offers substantial metabolic benefits, its long half-life and effects on gastric motility raise perioperative concerns, as delayed gastric emptying may persist in some patients even after 1 week of suspension[2,10,27].

Consistent with gastric ultrasonography reports, up to 90% of patients using GLP-1RAs have RGC despite ≥ 8 hours of fasting[1]. Definitions of increased RGC varied across studies, with thresholds ranging from 0.8 mL/kg to 1.5 mL/kg[2,9,10,11,26]. This definitional heterogeneity may have contributed to between-study variability.

Some studies have reported that a clear liquid diet in the 24 hours before procedures may help reduce RGC in patients using GLP1-RAs[1,8]. A study found same-day colonoscopy to be protective against RGC (OR = 0.41, 95%CI: 0.23-0.73), likely reflecting the effect of prolonged liquid fasting protocols[25]. Other observational studies have reported similar protective association, with one showing an OR of 0.34 (95%CI: 0.23-0.52)[1] and another demonstrating a further reduction when upper endoscopy was combined with colonoscopy (prevalence ratio 0.25, 95%CI: 0.16-0.39) compared with upper endoscopy alone[10].

Pulmonary aspiration under anesthesia, although rare (estimated incidence 1 in 3000-7000 elective procedures), involves substantial morbidity and remains the leading cause of anesthesia-related mortality[11,28]. Nearly half of the patients required intensive or high-dependency care, over one-third required mechanical ventilation, and mortality reached 6%-7%. It accounted for up to 50% of anesthesia-related mortality and 17% of major airway events, with death or permanent severe injury documented in more than 50% of reported cases[29-31].

Given this biologically plausible mechanism and the catastrophic consequences of aspiration, even a modest increase in risk would be clinically significant[32,33]. Consequently, prospective studies investigating pulmonary aspiration as a primary outcome, requires an impractically large sample size to achieve sufficient statistical power[9].

To contextualize the rarity of aspiration events, the sample size required to detect a small absolute difference in risk was estimated, assuming a baseline incidence of 0.05%[23], a reduction to 0.03%, and two-sided α of 0.05. Under these assumptions, approximately 300000-400000 participants would be needed to achieve conventional power, whereas our meta-analysis included only 11022 semaglutide users and 38629 controls. This large discrepancy highlights that the available data are substantially underpowered for aspiration analysis and helps explain the wide CIs around our pooled estimate.

Although aspiration-focused trials are unlikely to be feasible because of the rarity of the outcome, randomized studies assessing intermediate endpoints, such as ultrasound-assessed RGC under different fasting or semaglutide-holding strategies, are needed. Current evidence therefore relies mainly on observational studies[6].

Although most included studies achieved high NOS scores, these ratings should be interpreted cautiously because all included studies were observational and, thus, remain vulnerable to residual confounding, selection bias, and unmeasured differences between groups. The NOS is a useful tool for assessing key domains such as selection, comparability, and outcome assessment[15], but high scores do not eliminate the fundamental limitations of nonrandomized designs. Accordingly, the available evidence should be interpreted as associative rather than causal, and pooling these studies does not overcome the underlying risk of bias.

The original protocol also included a dose–response/time-since-last-dose meta-analysis to examine the relationship between interruption duration and outcomes, but this was not feasible because interruption timing was inconsistently reported and rarely comparable across studies[2,10,26]. Only a minority of studies reported interruption intervals, and these definitions were heterogeneous[2,10,12,26]. Therefore, our data cannot determine whether a 7-day interruption is superior to shorter or longer intervals, and this approach should be prospectively tested rather than adopted as a fixed standard.

Substantial heterogeneity was observed in the pooled analysis for RGC and remained present across sensitivity analyses. Although exploratory subgroup analyses by study design and procedure type did not identify statistically robust between-group differences under the random-effects model, these analyses suggest that several factors may have contributed to between-study variability. These include differences in procedure type, baseline patient characteristics, definitions of RGC and digestive symptoms, semaglutide interruption intervals, and perioperative management strategies. Importantly, many of these variables were incompletely or inconsistently reported, which limited more granular subgroup and meta-regression analyses. Therefore, the pooled estimates should be interpreted cautiously, and future studies should use more standardized definitions and exposure reporting to improve comparability across cohorts.

This study has several limitations. First, important clinical data, such as the interval of semaglutide interruption, American Society of Anesthesiologists physical status, and perioperative fasting duration, were not consistently available across the included studies, which limits our ability to perform more granular subgroup analyses. Second, definitions of RGC and digestive symptoms varied between studies, contributing to heterogeneity. Third, the outcomes of greatest clinical concern - pulmonary aspiration and procedure discontinuation - were rarely reported, resulting in wide CIs and low certainty of evidence. Fourth, no randomized controlled trials were identified; all the included studies were observational, so residual confounding cannot be excluded, even though most studies scored highly on the NOS. Fifth, although studies enrolling broader GLP-1RA populations were only included when semaglutide-specific subgroup data could be isolated, differences in subgroup reporting and disaggregation across studies may still have affected comparability.

Finally, aspiration risk depends not only on gastric volume but also on whether full stomach precautions, such as rapid-sequence induction or gastric ultrasound, were applied, and this information was absent or inconsistently reported across included studies. Whether clinicians already aware of semaglutide-related risk selectively modified their practice also cannot be determined, which could have biased our estimates toward the null.

Consequently, our findings cannot establish that semaglutide increases RGC or aspiration, only that these events were observed more frequently in semaglutide users in the available cohorts. Although pooling such studies improves precision, it does not overcome the fundamental limitations of nonrandomized designs.

Future studies should systematically evaluate additional outcomes beyond the primary endpoints, including length of hospital stay, postoperative pain, 30-day readmission rates, and delayed recovery. Standardized definitions, validated measurement tools, weekly doses, therapy duration, and consistent follow-up intervals are essential to ensure comparability across studies and better characterize the broader effect of preoperative semaglutide on surgical recovery.

Existing guidelines address the perioperative management of GLP-1RA users differently. Multisociety, Brazilian, and European Society of Anaesthesiology and Intensive Care guidelines all endorse some combination of selective drug interruption, extended clear-fluid intake, and gastric ultrasonography, but acknowledge that the evidence base remains limited. A comparative summary of these recommendations is provided in Table 4[5,34,35].

Table 4 Comparative recommendations for perioperative management of glucagon-like peptide-1 receptor agonists.

Guideline/source	Suspension timing	Dietary measures	Use of gastric ultrasound	Risk stratification/other notes
Multisociety Clinical Practice Guidance (ASA and others, 2024)[5]	No universal discontinuation. If high-risk: Daily formulations - hold on day of surgery; weekly formulations - hold 7 days prior. Duration beyond these intervals unknown	Consider ≥ 24 hours liquid diet if concern for delayed emptying	May be used when retained gastric content is a concern	High-risk factors: Dose escalation, high doses, weekly formulations, active gastrointestinal symptoms, comorbid dysmotility. Avoid discontinuation only in obesity (bias concern)
Brazilian Position Statement (SBD/SBA/ABESO, 2025)[34]	7 days for long-acting; 1 day for short-acting, in patients at increased aspiration risk or within 12 weeks of dose escalation/instability	Liquid diet 24 hours before + 8-12 hours fasting	POCUS is recommended whenever available	Emphasis on selective suspension plus diet + ultrasound
ESAIC Guideline (2025)[35]	≥ 7 days for weekly; ≥ 14 days if for obesity. Hold daily formulations on day of surgery	24 hours clear-fluid diet	Strongly encouraged; treat all GLP-1RA users as “full stomach” if content present	More conservative: Notes 1-week hold may not normalize gastric emptying; RSI if urgent

ASA: American Society of Anesthesiologists; ABESO: Brazilian Association for the Study of Obesity and Metabolic Syndrome; ESAIC: European Society of Anaesthesiology and Intensive Care; POCUS: Point-of-care ultrasound; RSI: Rapid-sequence induction; SBA: Brazilian Society of Anesthesiology; SBD: Brazilian Diabetes Society.

CONCLUSION

Current observational evidence suggests that perioperative semaglutide treatment may be associated with increased RGC and digestive symptoms. However, RGC is a surrogate marker rather than a direct clinical endpoint, and its increase should not be interpreted as direct evidence of increased pulmonary aspiration risk. Evidence regarding pulmonary aspiration, procedure discontinuation, and gastroparesis remains uncertain because of sparse event data, wide CIs, and low to very low certainty of evidence. Given the observational design of the included studies, substantial heterogeneity, and absence of randomized trials, these findings should be interpreted cautiously. Prospective studies using standardized definitions, fasting protocols, and perioperative management strategies are needed to clarify the clinical significance of these associations.

ACKNOWLEDGEMENTS

Assistance with the article

The authors gratefully acknowledge the support of colleagues from the Department of Anesthesiology and the Department of Nephrology and Biostatistics, University of Caxias do Sul, for their valuable input and assistance during this project.

Prior presentation

Preliminary results from this study were presented in abstract form at the American Society of Anesthesiologists Annual Meeting 2025, held on October 11-15, 2025, in San Antonio, Texas (Abstract #7227 - https://www.abstractsonline.com/pp8/#!/21028/presentation/7227).

Guarantor statement

Vinicius Remus Ballotin is the guarantor of this work and had full access to all data, taking responsibility for the integrity of the data and accuracy of the analyses.