Residual gastric content after holding of glucagon-like peptide-1 receptor agonists before elective surgery: a cross-sectional study - the RESIDUAL study Authors: Charles Boudreau, Marjorie Couture, Nicolas Rousseau-Saine, Pascal Laferrière-Langlois, Étienne Roy-Renaud, Julien Burey, Meggie Raymond, André Denault, Louis Morisson Published: 2026 ABSTRACT AbstractIntroductionGlucagon-like peptide-1 receptor agonists (GLP-1 RA) are now increasingly used for glycemic control in type 2 diabetes mellitus as well as for weight loss. Their effect on gastric emptying may increase the risk of aspiration in the perioperative setting; unfortunately, the optimal schedule of perioperative holding of GLP-1 RA is unclear.MethodsWe conducted a cross-sectional prospective study at two university affiliated hospitals in Montreal, Canada, comparing the preoperative prevalence of increased residual gastric content (RGC) on gastric ultrasound (GUS) between two groups of fasted adults scheduled for elective surgery: (i) an exposure group of patients receiving a weekly injectable GLP-1 RA and discontinued for ≥ 7 days before surgery, regardless of indication; and (ii) a control group of patients not receiving any GLP-1 RA, preferentially with type 2 diabetes or with a body mass index ≥ 25 kg/m². The primary outcome was increased RGC, defined as the presence of solid content, thick liquid, or > 1.5 mL/kg of clear liquid. Confounding was addressed using overlap-weighting propensity-score adjustment.Results93 patients were included. The unadjusted prevalence of increased RGC was 24% (10/41) in the control group and 42% (22/52) in the exposure group. Solid content accounted for 80% (8/10) and 82% (18/22) of positive exams in the control group and the exposure group, respectively. In the unadjusted analysis, there was no statistically significant difference in increased RGC between groups (prevalence ratio, 1.73; 95% CI, 0.96 to 3.75). There was no statistically significant difference between groups after adjustment for confounding (adjusted prevalence ratio, 1.65; 95% CI, 0.7 to 3.7) with an absolute average treatment effect of 15.4% (95% CI, -10 to 35.6).ConclusionsThe prevalence of increased residual gastric content was 42% in patients who had discontinued weekly GLP-1 RA for at least 7 days, compared with 24% in non-exposed controls. Although this difference was not statistically significant, our study was underpowered. The findings therefore represent a clinically relevant but statistically inconclusive signal that the recommended 7-day discontinuation interval may not consistently restore an empty stomach in patients on weekly GLP-1 RA.Trial registrationThis study was registered on clinicaltrials.gov (July 8th, 2024, NCT06500143).Supplementary InformationThe online version contains supplementary material available at 10.1186/s12871-026-03999-2.Keywords: Ultrasonography, Glucagon-Like Peptide-1 Receptor Agonists, Diabetes Mellitus, Respiratory Aspiration of Gastric Contents, Perioperative care, Fasting Background Glucagon-like peptide-1 receptor agonists (GLP-1 RA) are glucose-lowering drugs that act through three mechanisms: stimulation of insulin secretion, inhibition of glucagon secretion and decrease of gastric emptying [1]. GLP-1 RA are now an important part of glycemic control in type 2 diabetes mellitus (T2DM) and have also been approved for weight loss [1]. GLP-1 RA have been shown to increase the risk of pancreatitis, biliary disease, bowel obstruction occlusion and gastroparesis [2]. The clinical implication of the latter in the periprocedural setting is uncertain but endoscopic studies have shown increased residual gastric content (RGC) with GLP-1 RA use [3, 4]. Gastric ultrasound (GUS) is a validated bedside method of assessing gastric content and identifying patients at increased risk of aspiration in the perioperative setting [5]. Several recent ultrasound-based studies have shown an association between GLP-1 RA exposure and increased RGC in fasted surgical patients [6, 7]. However, in most of these reports the median time since last GLP-1 RA dose was substantially shorter than the 7-day interval now recommended by multisociety guidance for weekly formulations [8, 9], and patients who had complied with this recommendation were either excluded or represented only a small subgroup. Whether the recommended ≥ 7-day interruption restores an empty stomach in patients on weekly GLP-1 RA is therefore not yet established. Pulmonary aspiration of gastric content, although infrequent (approximately 1 in 4,000 to 10,000 fasted elective procedures [8, 10]) can lead to pneumonitis, pneumonia, pulmonary obstruction, brain damage and death [8, 11]. Recently published multisociety clinical guidelines on the use of GLP-1 RA in the perioperative period highlight the need to assess individual variables associated with increased risk of aspiration such as weekly dosing of GLP-1 RA, higher dose, presence of gastrointestinal symptoms as well as medical conditions that could delay gastric emptying [9]. If an elevated risk of aspiration exists, the GLP-1 RA may be withheld seven days before surgery for weekly formulations. Considering the extended half-lives of once-weekly GLP-1 RA (approximately 6 days for semaglutide and 5 days for dulaglutide), it is unclear if following those guidelines is sufficient to ensure patients are not at high risk of aspiration [12]. Recent case-reports and editorials have raised concerns about aspiration with GLP-1 RA even with appropriate holding [12–14]. In a recent prospective observational cross-sectional study, GLP-1 RA users had an increased RGC prevalence using GUS compared with non-users in a perioperative setting [6]. Most patients had received their last dose within 5 days of the procedure, thus not complying with the recent recommendations from the American Society of Anesthesiologists (ASA). In light of the paucity of evidence regarding the appropriate timing of GLP-1 RA holding before elective surgery, we chose to conduct a study comparing ultrasound-assessed RGC between once-weekly GLP-1 RA users that stopped their medication for 7 days or more before elective surgery and patients not taking GLP-1 RA. We hypothesized that even with appropriate holding of GLP-1 RA, those patients may have increased RGC and be at higher risk of aspiration. Secondarily, we aimed to explore the relation between the dose, as well as the number of days since the last dose of GLP-1 RA, and increased RGC. Methods We conducted a prospective cross-sectional study. Participants were prospectively recruited from October 2024 through February 2025 at two university-affiliated hospitals in Montreal, Quebec, Canada. The study received approval from our local IRB on August, 14th 2024 (Comité d’éthique et de la recherche du CIUSSS de l’Est de l’Ile de Montréal, IRB number: MP-12-2025-3794). All procedures involving human participants were performed in accordance with the ethical standards of the institutional research committee and with the 1964 Declaration of Helsinki and its later amendments. Written informed consent was obtained from every participant before enrollment. We recruited patients scheduled for elective surgery under anesthetic care in the operating room. Eligible patients had to be at least 18 years old with an ASA physical status classification of I to IV and appropriate fasting according to current Canadian guidelines: 6 h for a meal or solid food, and 2 h for clear liquid [15]. We excluded patients with history of large hiatal hernia or gastric surgery, as gastric ultrasound is not validated in this specific population [16]. We also excluded patients with pregnancy or unable to lay in right lateral decubitus. At our institutions, patients using weekly GLP-1 RA agonists are screened at the preoperative clinic and instructed to stop their medication 7 days before their surgery. Moreover, GLP-1 RA usage as well as presence of diabetes are available on the daily surgical schedule in the operating room management software. Using this software, a dedicated nurse screened the patients and recruited them by phone the day before or in person the morning of their elective surgery. Each participant was given informed consent and agreed to participate in the study. Exposure was defined as weekly injectable semaglutide, tirzepatide or dulaglutide use for more than 4 weeks, regardless of the therapeutic indication, with the last dose at least 7 days before surgery. Because the escalation phase is associated with a higher risk of delayed gastric emptying, we restricted the eligibility to patients with a duration of treatment of more than 4 weeks to mitigate the risk of overestimating the effect in the GLP-1 RA group [9]. We collected information on the specific drug, dose (in milligrams) and day of last injection. Patients in the control group were eligible patients who were not receiving any GLP-1 RA. Because GLP-1 RA prescription in the province of Quebec is overwhelmingly tied to type 2 diabetes, an exclusively non-diabetic control group would have created severe confounding by indication and would not have been logistically feasible within our recruitment window. We therefore recruited diabetic patients as controls, except when a non-diabetic patient was included in the exposure group; in that case, a matched non-diabetic overweight (BMI ≥ 25 kg/m²) control was subsequently enrolled. All data was collected using a questionnaire and available electronic medical records. We used REDCap software (Vanderbilt University) to record the data. We collected information on age, sex, body mass index (BMI), diabetes, hypertension, coronary and peripheral artery disease, smoking, gastroesophageal reflux disease (GERD), ASA classification, pre-operative pain, gastroparesis, chronic pain, opioid and proton pump inhibitor (PPI) use and time of last solid and liquid intake. Glomerular filtration rate (GFR) was calculated using creatinine level and the CKD-EPI formula. Each specific surgery procedure was documented and classified according to the surgical specialty. Preoperative pain was assessed using a verbal numerical rating scale of 0 (no pain) to 10 (severe pain). Gastroparesis was defined as active treatment with a prokinetic drug. All information was collected by a dedicated nurse and/or a member of the research team. To avoid bias before assessment of gastric content, the ultrasound operator could collect patient information only if the result of the ultrasound was beforehand recorded in the database. The primary outcome was increased RGC on GUS defined by the presence of thick liquid, solid or more than 1.5 mL.kg-1 of clear liquid content. The GUS was performed by trained anesthesiologists who were not in charge of the patient (see Supplemental digital content for description of the technique). After GUS, result of the exam was immediately recorded in the database and disclosed to both the patient and the anesthesiologist caring for the patient. A second trained anesthesiologist blinded to the initial interpretation subsequently reviewed the saved images. In case of discrepancy, a third blinded trained anesthesiologist made the final interpretation. Ultrasound operators and reviewers were blinded to the group assignment and did not participate in patients screening. Demographics and covariates are presented using median with interquartile range (IQR) for continuous variables and absolute count and frequency (%) for categorical variables. We assessed the baseline imbalance between groups using absolute standardized differences (ASD) [17]. For the primary analysis of increased RGC (dichotomous), we estimated an unadjusted prevalence ratio (PR) using a generalized linear model with a log link and Poisson distribution with robust error variance [18]. To adjust for confounding variables, we planned to use an augmented inverse probability weighting (AIPW) estimator, a propensity score-based method [19]. The data analysis and statistical plan was written and filed with our institutional review board before data were accessed. However, because of inadequate covariates balancing using the AIPW, we used an overlap weighting (OW) technique [20]. The OW method results in exact mean balance of covariates. The propensity score for each patient was estimated using a logistic regression model adjusted for age, sex, ASA classification, BMI, diabetes, moderate to severe preoperative pain (> 3 on the verbal rating scale), opioid use and time since last oral intake. For adjustment, we chose covariates based on available literature [16, 21, 22]. We deliberately did not adjust for GERD, gastroparesis (active prokinetic treatment) or PPI use, because each of these variables can plausibly act as a mediator or collider on the causal pathway from GLP-1 RA exposure to increased RGC: GLP-1 RA-induced gastrointestinal symptoms can lead to gastroparesis diagnoses, prokinetic prescription and PPI initiation, and these in turn affect gastric emptying. Including such variables in a propensity-score model would have opened non-causal pathways and biased the estimated effect of GLP-1 RA on gastric content [2, 23]. After overlap weighting, we computed weighted frequencies (categorical variables) and weighted medians (continuous variables) and re-assessed balance using ASD; an ASD ≥ 0.10 was considered to indicate clinically meaningful imbalance [17]. The OW was then used to estimate the average treatment effect (ATE) — the absolute difference in the prevalence of increased RGC between groups — and the adjusted PR. We used non-parametric bootstrapping with 1,000 resamples (with replacement) to compute percentile-based 95% confidence intervals for the ATE and the adjusted PR [24]. We performed pre-specified sensitivity analyses by repeating the primary analysis using multivariable logistic regression and AIPW. For the secondary analyses, we investigated the association between the time since last injection of GLP-1 RA as well as the time since the last oral intake and the prevalence of increased RGC using a logistic regression model. The same covariates included in the primary analysis were used for adjustment. For the patients receiving semaglutide, we investigated the association between the dose of semaglutide and the prevalence of increased RGC using the same strategy. These exploratory analyses should be considered hypothesis-generating only and interpreted with caution given the limited sample size and the small number of events. All statistical analyses were performed using R programming software using two-sided tests and a threshold of p < 0.05. Based on previous ultrasound-based studies in mostly non-diabetic elective surgical populations, we estimated a baseline prevalence of increased RGC of 10% in the control group [3, 6, 25, 26]. To detect a 30-percentage-point absolute increase in increased RGC in patients exposed to weekly semaglutide with 90% power and a two-sided alpha of 0.05, we calculated that 84 subjects would be required (42 per group). To account for withdrawal or loss of 10%, the final target sample size was 94 subjects (47 per group). Results A total of 94 patients undergoing elective procedures were included in our study. One patient in the GLP-1 RA group was excluded after undergoing GUS because it was ultimately determined that the last GLP-1 RA dose was less than 7 days before surgery. Of the remaining 93 patients, 52 (56%) were receiving GLP-1 RA (exposure group) and 41 (44%) were not (control group). The full details of enrollment are presented in Supplementary Figure S4. Baseline characteristics of patients in both groups as well as crude ASD are presented in Table 1. Adjusted ASD for covariates used for confounding adjustment are available in Table 1. In the exposure group, 50/52 (96%) of patients were receiving semaglutide, compared to one patient each for tirzepatide and dulaglutide. The main indication for GLP-1 RA treatment was diabetes (45/52; 87%) and the median time since last dose was 11 days (IQR 9 to 13). There was no missing data for any of the patients in our study. Table 1– Baseline characteristics of patientsCharacteristicControl group* (n = 41)GLP-1 RA group* (n = 52)ASD (unadjusted)Age, median [IQR]71 [67, 78]65 [58, 72]0.622Sex Male26 (63.4)29 (55.8)0.156 Female15 (36.6)23 (44.2) BMI, median [IQR]29.0 [27.0, 35.0]33.0 [27.0, 38.2]0.377Admission type Hospitalization32 (78.035 (67.3)0.243 Ambulatory9 (22.0)17 (32.7)Surgical specialty General5 (12.2)11 (21.2)0.625 Orthopedic8 (19.5)13 (25.0) Cardiac9 (22.0)7 (13.5) Urology9 (22.0)4 (7.7) Thoracic3 (7.3)5 (9.6) Gynecology3 (7.3)1 (1.9) ENT2 (4.9)5 (9.6) Other2 (4.9)6 (11.5)ASA physical status class I-II17 (41.5)25 (48.1)0.228 III15 (36.6)20 (38.5) IV9 (22.0)7 (13.5)Tobacco use Active5 (12.2)7 (13.5)0.091 Discontinued20 (48.8)23 (44.2) Hypertension38 (92.7)42 (80.8)0.357 Peripheral artery disease9 (22.0)9 (17.3)0.117 Coronary artery disease20 (48.8)17 (32.7)0.332 Diabetes36 (87.8)47 (90.4)0.083 GERD10 (24.4)22 (42.3)0.387 Gastroparesis0 (0.0)2 (3.8)0.283 PPI use13 (31.7)19 (36.5)0.102 Chronic pain12 (29.3)25 (48.1)0.394 Opioid use0 (0.0)8 (15.4)0.603Preoperative pain† None32 (78.0)29 (55.8)0.529 Mild-Moderate6 (14.6)19 (36.5) Severe3 (7.3)4 (7.7) Estimated_GFR, median [IQR]75.0 [48.0, 94.0]79.0 [63.2, 91.8]0.162 Time since last solid intake, median [IQR]14.0 [13.0, 16.0]13.5 [11.4, 15.0]0.222 Time since last liquid intake, median [IQR]3.0 [2.2, 4.0]3.0 [2.0, 5.8]0.253* All prevalences are reported as number (percentage) unless specified† Pain was evaluated with a verbal numerical rating questionnaire (No pain = 0, mild to moderate pain < 7 and severe pain ≥ 7) – Baseline characteristics of patients * All prevalences are reported as number (percentage) unless specified † Pain was evaluated with a verbal numerical rating questionnaire (No pain = 0, mild to moderate pain < 7 and severe pain ≥ 7) In our primary analysis, the unadjusted prevalence of increased RGC was 24% (10/41) in the control group and 42% (22/52) in the exposure group, corresponding to an unadjusted prevalence ratio of 1.73 (95% CI 0.96 to 3.75). After overlap-weighting adjustment for age, sex, ASA physical status, BMI, diabetes, opioid use, moderate-to-severe preoperative pain and time since last oral intake, the adjusted prevalence ratio was 1.65 (95% CI 0.70 to 3.70) and the adjusted average treatment effect was an absolute difference of 15.4% points (95% CI − 10.0 to 35.6) (Table 2). Table 2Comparison of confounding covariates between groups after overlap weighting (OW) analysisCharacteristicControl group (n = 17.57)*GLP-1 RA group (n = 17.57)*ASD (adjusted)Age, median [IQR]69 [65, 75]69 [60, 75]< 0.001Female sex6.7 (37.9)6.7 (37.9)< 0.001BMI, median [IQR]31.18 [27.00, 35.00]32.00 [26.23, 35.00]< 0.001ASA physical status class I-II8.6 (48.9)8.6 (48.9)< 0.001 III6.6 (37.4)6.6 (37.4) IV2.4 (13.7)2.4 (13.7) Diabetes15.5 (88.1)15.5 (88.1)< 0.001 Opioid use0.0 (0.0)0.0 (0.0)< 0.001 Preoperative pain†3.4 (19.3)3.4 (19.4)< 0.001 Time since last oral intake, median [IQR]2.50 [2.15, 4.00]2.50 [2.00, 4.62]< 0.001* All prevalences are reported as number (percentage) unless specified† Preoperative pain was considered significant is the verbal rating scale was more than 3Abbreviations: ASA, American Society of Anesthesiologists; ASD, absolute standard difference; GERD, gastroesophageal reflux disease; GLP-1 RA, glucagon-like peptide-1 receptor agonistASD denotes the absolute standardised difference; an ASD ≥ 0.10 was used to indicate clinically meaningful imbalance between groups [17] Comparison of confounding covariates between groups after overlap weighting (OW) analysis * All prevalences are reported as number (percentage) unless specified † Preoperative pain was considered significant is the verbal rating scale was more than 3 Abbreviations: ASA, American Society of Anesthesiologists; ASD, absolute standard difference; GERD, gastroesophageal reflux disease; GLP-1 RA, glucagon-like peptide-1 receptor agonist ASD denotes the absolute standardised difference; an ASD ≥ 0.10 was used to indicate clinically meaningful imbalance between groups [17] Solid content accounted for the majority of positive examinations in both groups: 8/10 (80%) of positive scans in the control group and 18/22 (82%) in the exposure group, with the remainder consisting of thick liquid (2 controls, 3 exposed) and clear liquid above the 1.5 mL/kg threshold (0 controls, 1 exposed) (Table 3). Table 3– Association between glucagon-like peptide-1 receptor agonist (GLP-1 RA) use and increased residual gastric contentUnadjustedAdjusted*Sensitivity analysis †Prevalence ratio (95% CI)1.73 (0.96 to 3.75)1.65 (0.7 to 3.7)1.84 (0.74 to 4.24)Average treatment effect, % (95% CI)NA15.4 (-10 to 35.6)18.1 (-10.2 to 38.7)* An overlap weighting (OW) technique was used to adjust for age, sex, American Society of Anesthesiologists physical status classification, BMI, diabetes, opioid use, moderate to severe preoperative pain (> 3 on the verbal rating scale) and time since last oral intake† An augmented inverse probability weighting (AIPW) matching technique was used to adjust for the same covariates as the primary analysis – Association between glucagon-like peptide-1 receptor agonist (GLP-1 RA) use and increased residual gastric content * An overlap weighting (OW) technique was used to adjust for age, sex, American Society of Anesthesiologists physical status classification, BMI, diabetes, opioid use, moderate to severe preoperative pain (> 3 on the verbal rating scale) and time since last oral intake † An augmented inverse probability weighting (AIPW) matching technique was used to adjust for the same covariates as the primary analysis Table 4– Prevalence of increased residual gastric contentGUS resultControl group (n = 41)GLP-1 RA group (n = 52)Empty31 (75.6)30 (57.7)Increased RGC10 (24.4)22 (42.3) Type of increased RGC Clear liquid0 (0)1 (4.5)Thick liquid2 (20.0)3 (13.6)Solid8 (80.0)18 (81.8)* All prevalences are reported as number (percentage) unless specifiedAbbreviations: GLP-1 RA, glucagon-like peptide-1 receptor agonist; GUS, gastric ultrasound; RGC, residual gastric content – Prevalence of increased residual gastric content * All prevalences are reported as number (percentage) unless specified Abbreviations: GLP-1 RA, glucagon-like peptide-1 receptor agonist; GUS, gastric ultrasound; RGC, residual gastric content In our secondary analyses, there was no statistically significant association between the time since last GLP-1 RA injection and increased RGC (adjusted odds ratio per additional day, 1.08; 95% CI 0.86 to 1.39; Supplementary Table S1 and Supplementary Figure S8 and S9), nor between the time since last oral intake and increased RGC (adjusted odds ratio per hour, 1.12; 95% CI 0.9 to 1.46; Supplementary Table S1 and Supplementary Figure S10). Finally, we observed no statistically significant association between the dose of semaglutide and increased RGC (adjusted odds ratio per mg, 2.19; 95% CI 0.47 to 12.7; Supplementary Table S2 and Supplementary Figure S11). The primary results were confirmed in our sensitivity analyses using the AIPW, where GLP-1 RA treatment was not associated with a higher risk of increased RGC. We also conducted a multivariate logistic regression that did not show any statistically significant difference for the primary outcome. Discussion In this prospective, blinded, propensity-score-adjusted cross-sectional study of fasted adults scheduled for elective surgery, the prevalence of increased residual gastric content on gastric ultrasound was 42% in patients who had discontinued a weekly GLP-1 RA for at least 7 days, compared with 24% in non-exposed controls. This 18% absolute difference (adjusted ATE 15.4%, 95% CI − 10.0 to 35.6) was not statistically significant; the wide confidence interval and the higher-than-expected control prevalence suggest that this finding most likely reflects insufficient power rather than absence of effect. The use of overlap weighting allowed effective balancing of all measured baseline characteristics between two clinically distinct groups (all adjusted ASD < 0.001), supporting the internal validity of the findings despite the underlying population heterogeneity. Neither time since last injection (median 11 days) nor dose was associated with the prevalence of increased RGC. Finally, we observed no association between the time since last oral intake and increased RGC. These secondary analyses should be considered exploratory and interpreted with caution given the limited sample size. These findings are consistent with previous studies demonstrating increased residual gastric content in GLP-1 RA users, including those using endoscopic assessment [4], which together suggest that shorter interruption intervals may be insufficient to normalise gastric emptying in a meaningful proportion of patients. Our results extend this evidence to the specific subgroup of patients who have complied with the currently recommended ≥ 7-day discontinuation interval for weekly formulations. In a recent cross-sectional prospective trial of patients undergoing elective procedures, Sen et al. [6] reported a prevalence of increased RGC on gastric ultrasound of 56% (35/62) in weekly GLP-1 RA users compared with 19% (12/62) in non-users; Nersessian et al. [7] have reported broadly similar findings. The smaller crude prevalence of increased RGC in the GLP-1 RA group in our study compared with Sen et al. is most plausibly explained by the longer median time since last injection in our cohort (11 days versus approximately 5 days), suggesting a partial — but incomplete — recovery of gastric emptying with longer interruption. The relatively high prevalence of increased RGC in the control group (24%) is most plausibly explained by the high proportion of patients with type 2 diabetes (88%), a population known to have delayed gastric emptying independently of GLP-1 RA exposure. This composition reflects the real-world target population for perioperative aspiration risk assessment in GLP-1 RA users, but also reduced the contrast between groups and likely biased our comparison toward the null. Notably, our control prevalence is closer to the 19% reported by Sen et al. in their similarly diabetes-enriched control group than to that reported in mostly non-diabetic elective surgical cohorts [6, 21, 26, 27], suggesting that diabetic patients may merit gastric-content screening even in the absence of GLP-1 RA exposure. In both groups, solid content accounted for the great majority of positive examinations (8/10, 80% in controls; 18/22, 82% in the exposure group), pointing to delayed solid emptying — rather than residual liquid — as the dominant mechanism of incomplete fasting in this population; however, our cross-sectional design does not support specific recommendations regarding preoperative diet composition. We did not detect a monotonic dose-response or duration-response relationship between GLP-1 RA exposure and increased RGC. Several factors may contribute to this absence of signal: (i) the limited number of events restricted statistical power for trend testing; (ii) variability in time since last injection and in semaglutide dose was modest in our cohort; and (iii) inter-individual differences in semaglutide pharmacokinetics, pharmacogenomics and gastric emptying may dominate over dose- or interval-related effects at the doses used in clinical practice. In our exploratory analysis, we observed no association between fasting duration and increased RGC; this is most likely explained by the limited variability of fasting times in our cohort, since the institutional standard of fasting from midnight produced narrow distributions in both groups (median 14 h, IQR 13–16 in controls; 13.5 h, IQR 11.4–15 in exposed), restricting our ability to detect a dose-response relationship. Because 50 of our 52 exposed patients were receiving weekly semaglutide, our findings primarily apply to weekly semaglutide formulations rather than to GLP-1 receptor agonists as a class. Extension to weekly dulaglutide and tirzepatide is biologically plausible on pharmacokinetic grounds but should be confirmed empirically. Our results cannot be extrapolated to daily formulations or to oral semaglutide. The findings of this study, in conjunction with emerging evidence, raise important questions regarding the adequacy of currently recommended interruption intervals (e.g., 7 days for weekly GLP-1 RA formulations as suggested by the American Society of Anesthesiologists). Although the present study is not powered to establish definitive thresholds, the observed trend toward increased residual gastric content despite ≥ 7-day interruption suggests that this interval may be insufficient in some patients. Future studies should aim to better define optimal discontinuation intervals and may help inform potential refinements of current perioperative recommendations.” Limitations Our study has several limitations. The most important limitation is structural: the composition of our control group itself biased the comparison toward the null. By design, our controls carried a high intrinsic risk of delayed gastric emptying because of the strong overlap between diabetes and GLP-1 RA prescription in our region. As a result, the contrast between groups was attenuated and the apparent GLP-1 RA effect underestimated. We consider this a structural feature of any single-centre study of perioperative GLP-1 RA in regions where diabetes is the dominant indication for these agents. Despite balancing on age, sex, BMI, ASA class, diabetes status, opioid use, preoperative pain and fasting duration, residual confounding by unmeasured diabetes-related factors — including duration of diabetes, glycaemic control (HbA1c), and presence of autonomic neuropathy — cannot be excluded. These factors are plausible determinants of gastric emptying and may have differed between groups even after overlap weighting. This is an important limitation of any propensity-score adjustment based on routinely available covariates. Patient screening relied on the data available on surgical schedules. Patients may have been missed because of incomplete information on these lists (e.g., undocumented diabetes or GLP-1 RA use [28]; a recent case report describes intraoperative aspiration in a patient who did not disclose semaglutide use during preoperative evaluation, illustrating the consequences of incomplete disclosure). Exhaustive chart review of every scheduled patient was not feasible, given that — to maintain blinding of the GUS operators — only the dedicated research nurse was permitted to participate in screening. There were imbalances between the groups, addressed using robust propensity-score weighting; nevertheless, a randomised controlled trial would have been infeasible because it would imply initiating GLP-1 RA weeks before a planned date of surgery, which is generally scheduled only one to two weeks in advance at our institutions. The dominant association of GLP-1 RA prescription with diabetes also explains why the two groups were unbalanced (41 controls vs. 52 exposed) despite reaching our overall sample-size target. We used increased RGC on gastric ultrasound as a surrogate for increased aspiration risk, using a validated method and validated criteria [5, 16]. The available literature does not allow precise estimation of how a given level of RGC translates into aspiration risk in fasted patients [29]. Furthermore, because pulmonary aspiration during elective procedures is rare (1 in 3,000 to 10,000), a trial powered to detect differences in aspiration itself is unlikely to be undertaken [11]. Increased RGC on GUS therefore remains the most accessible and reliable bedside surrogate for individualising aspiration risk assessment in this population. Patients identified as high-risk should be managed with evidence-based aspiration-mitigating techniques such as rapid-sequence intubation or regional anaesthesia with minimal sedation [9, 10]. Although the gastric ultrasound result was determined and recorded by a blinded operator before being communicated to the anaesthetist in charge — protecting the primary outcome from interpretation bias — disclosure of the result may have influenced subsequent anaesthetic management (for example, the choice of rapid-sequence induction or postponement of surgery). This could have attenuated any signal in downstream clinical outcomes such as aspiration; however, we did not collect such outcomes in the present study. Finally, our sample-size calculation assumed a baseline prevalence of 10% increased RGC in the control group, derived from previous ultrasound studies in elective surgical populations comprising mostly non-diabetic patients [3, 21, 26]. We acknowledge that this assumption likely underestimated the true baseline prevalence in our diabetes-enriched control group; the higher-than-expected observed prevalence (24%) and the smaller-than-anticipated effect size mean that our study is most plausibly underpowered for the population actually recruited, with a corresponding increased risk of type II error. Using the observed values (24% control prevalence; 18% absolute difference), approximately 212 participants per group would have been required to achieve 90% power. The lack of statistical significance should therefore be interpreted with caution. Conclusions Our findings suggest that the recommended 7-day discontinuation interval for weekly GLP-1 RA, in particular weekly semaglutide, may not reliably restore an empty stomach. We observed an increased residual gastric content in 42% of exposed patients despite adequate discontinuation, compared with 24% in non-exposed controls. Until adequately powered prospective studies become available, individualized perioperative aspiration-risk assessment — potentially including bedside gastric ultrasound in selected patients — remains a reasonable approach in patients receiving weekly GLP-1 RA. Supplementary Information Supplementary Material 1. Acknowledgements Not applicable. Authors’ contributions Charles Boudreau contributed to conception and design, acquisition of data, analysis and interpretation of date, drafting and revising the article and final approval of the version to be published.Marjorie Couture contributed to conception and design, acquisition of data, analysis and interpretation of data, and final approval of the version to be published.Nicolas Rousseau-Saine contributed to acquisition of data, revising the article and final approval of the version to be published.Pascal Laferrière-Langlois contributed to conception and design, revising the article and final approval of the version to be published.Étienne Roy-Renaud contributed to acquisition of data, revising the article and final approval of the version to be published.Julien Burey contributed to conception and design, revising the article and final approval of the version to be published.Meggie Raymond contributed to acquisition of data, revising the article and final approval of the version to be published.André Denault contributed to conception and design, acquisition of data, revising the article and final approval of the version to be published.Louis Morisson contributed to conception and design, acquisition of data, analysis and interpretation of date, drafting and revising the article, and final approval of the version to be published. Funding Support was provided solely from institutional and/or department sources. Data availability The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The study received approval from our local IRB on August, 14th 2024 (Comité d’éthique et de la recherche du CIUSSS de l’Est de l’Ile de Montréal, IRB number: MP-12-2025-3794). All procedures involving human participants were performed in accordance with the ethical standards of the institutional research committee and with the 1964 Declaration of Helsinki and its later amendments. All participants signed the informed consent form before participation to the study. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.