Glucagon-like peptide-1 receptor agonists and gastric emptying time: a systematic review and meta-analysis of prospective studies.

Methods

This review followed the Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) reporting guidelines.12 We registered this review in the International Prospective Register of Systematic Reviews (PROSPERO; CRD42023461665) on 8 September 2023. An information specialist (M. E.) conducted systematic literature searches on 12 July 2023, using the Ovid platform for the databases MEDLINE, MEDLINE In-Process/ePubs, Embase, Cochrane Central Register of Controlled Trials, American Psychological Association (APA) PsycINFO®, and Cumulative Index to Nursing and Allied Health Literature (CINAHL). Title and abstract screening started after registration on 15 September 2023. We performed a full search update on 29 July 2024 and a supplemental search including titles until 30 December 2024 using citation search on Google Scholar, PubMed®, and the reference lists of included studies to identify any additional articles.

We performed a preliminary search and mined the selected full-text articles for potential keywords and appropriate controlled vocabulary terms (such as Medical Subject Headings [MeSH] for MEDLINE and Emtree descriptors for Embase). We used the Yale MeSH Analyzer to assess target citations.13 We built search strategy concept blocks and fleshed out each component with controlled vocabularies, text word terms, and synonyms. Electronic Supplementary Material (ESM) eTable 1 provides the complete Ovid MEDLINE search strategy. We limited the search to original articles that were in indexed journals, were in the English language, were about humans, and studied adults aged ≥ 18 yr. We excluded research-in-progress reports, case reports, study protocols, conference abstracts, theses/dissertations, retrospective studies, and non-English publications.

Two independent reviewers (A. C. and K. Z.) carried out title and abstract screening using Covidence systematic review management software (Melbourne, VIC, Australia). These two investigators performed full-text reviews of the selected articles, and conflicts were resolved through a third reviewer (F. C.). Inclusion criteria were 1) patients aged ≥ 18 yr taking a GLP-1 RA for diabetes mellitus and/or weight loss, 2) gastric emptying assessment reported as emptying T½, and 3) study design that was a randomized controlled trial (RCT) or a prospective cohort study. All modalities of gastric emptying assessment reporting T½, such as scintigraphy, stable isotope breath testing, and acetaminophen-based absorption testing, were included. All GLP-1 RA formulations were included.

Two independent reviewers (A. C. and K. Z.) extracted data from the selected studies using a spreadsheet designed a priori. They extracted the following variables: author, country, year of publication, study design, indication for GLP-1 RA use, sample size, age, sex, body mass index (BMI), type and dose of GLP-1 RA, length of study, gastric emptying test, baseline and posttreatment gastric emptying T½. We converted median [interquartile range (IQR)] values to mean (standard deviation [SD]) where possible.

Two independent reviewers (A. C. and K. Z.) evaluated study quality based on the study design. A third independent reviewer (F. C.) resolved disagreements. We used the revised Cochrane Risk of Bias 2 tool (RoB2) for RCTs for the primary outcome of gastric emptying T½.14 The RoB2 assesses bias arising from the following domains: randomization process, deviations from the intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. For nonrandomized cohort studies, we evaluated quality using the Risk of Bias in Non-randomized Studies—of Interventions (ROBINS-I) tool (ESM eFigs. 1 and 2).15 The domains assessed included confounding, selection of participants, classification, missing outcome data, measurement of outcomes, and selection of the reported result. For each domain in either tool, we rated the risk of bias as low, intermediate, or high on the basis of the methods used to minimize each source of bias. The overall risk of bias corresponded to the worst risk of bias in any of the domains (ESM eFig. 1).

We performed an assessment of the certainty of the evidence with the use of the Grading of Recommendations Assessment, Development and Evaluation (GRADE) classification.16

The primary analysis focused on the comparison of gastric emptying T½ between an active treatment group vs placebo (for RCTs) or between baseline and posttreatment values for prospective cohort studies. The preferred approach for calculating the effect size was standardized mean difference (SMD). Pim Cuijpers et al. generally advise against using within-group effect sizes in meta-analyses.17 In general, when data from both experimental and control groups are available, it is more reliable to compute between-group SMDs rather than pre–post comparisons.17 Nevertheless, SMDs may be biased in studies with small sample sizes (n ≤ 20),18 possibly leading to an overestimation of the true effect size. To address this, we corrected SMDs for small-sample bias, producing an effect size known as Hedges’ g.18 In addition, to make the results more clinically meaningful, we back-transformed the results to be expressed as a mean prolongation of gastric emptying T½ in min.

To address within-study heterogeneity, we conducted a three-level meta-analysis. The three levels of meta-analysis were as follows: 1) The first level represents the individual participants considered by each study author; 2) the second level reflects the clustered results reported within each study; and 3) the third level accounts for variability across different studies. Glucagon-like peptide-1 receptor agonists included in the subgroup analysis were 1) short-acting agents (exenatide, lixisenatide) and 2) long-acting agents (liraglutide, semaglutide). We evaluated heterogeneity by I2. To assess the sources of heterogeneity, we analyzed the short/long-acting subgroups using meta-regression, adjusted for study duration (in days), sample size, and publication year. In addition, we examined interactions to investigate whether the relationships between study duration, sample size, publication year, and effect size differed between subgroups. The effect of treatment duration was studied descriptively within three groups: 1) 1–3 weeks, 2) 4–10 weeks, and 3) > 10 weeks. To visually assess the effect of the publication year, we generated a cumulative forest plot. We converted data initially reported as median [IQR] to mean (SD) using validated methodologies wherever feasible.19,20 We used a funnel plot to assess publication bias. We used R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria) along with the packages “metafor” and “meta” to carry out the analyses. We considered a P < 0.05 to be significant. We present the results of meta-analysis in the form of forest plots.

Results

We summarized the results of the literature search in a PRISMA Flow Diagram (Fig. 1). We identified and screened 5,835 titles and abstracts. After removing duplicate studies and assessing titles and abstracts for inclusion criteria, we selected 132 articles for full-text reviews, following which we excluded 122 studies owing to the reasons listed in Fig. 1. Ten articles (total N = 300) fulfilled the inclusion criteria and were included in the meta-analysis.21–30 Table 1 summarizes the study characteristics. Eight studies were RCTs,21–25,27–29 while the remaining two were prospective cohort studies.26,30Fig. 1PRISMA flow diagramGLP-1 RA = glucagon-like peptide-1 receptor agonist; PRISMA = Preferred Reporting Items for Systematic Review and Meta-analysesTable 1Demographics and characteristics of studiesStudy, countryStudy designIndication for GLP-1 RA useSample size (N)Percentage maleAge (yr), mean (SD)BMI (kg·m−2), mean (SD)GLP-1 RADoseControl treatment*TestDuration of studyShort-actingLinnebjerg 2008, USACross-over three-arm studyT2DM1794%57 (10)30 (4)Exenatide5 µg bid10 µg bidPlaceboScintigraphy5 daysLorenz 2013, GermanyRCTT2DM2152%54 (7)31 (4)Lixisenatide20 µg qdPlacebo¹³C-octanoic acid breath test4 weeksBeti 2018, GermanyPCT2DM2065%51 (12)42 (10)Exenatide5 µg bidN/A¹³C-octanoic acid breath test4 daysMeier 2019, GermanyRCTT2DM2891%60 (9)31 (4)Lixisenatide20 µg qdInsulin glargine¹³C-octanoic acid breath test4 weeksLong-actingNagai 2014, JapanPCT2DM1657%47 (12)36 (7)Liraglutide0.9 mg qdN/A¹³C-octanoic acid breath test7 daysDejgaard 2016, DenmarkRCTT1DM5060%47 (13)30 (4)Liraglutide1.8 mg qdPlaceboAcetaminophen absorption3, 24 weeksHalawi 2017, USARCTObesity19NR42 (14)37 (6)Liraglutide3.0 mg qdPlaceboScintigraphy5, 16 weeksQuast 2021, GermanyRCTT2DM5064%60 (7)31 (1)LiraglutideLixisenatide1.8 mg qd20 µg qdN/A¹³C-octanoic acid breath test10 weeksMaselli 2022, USARCTObesity5969%42 (14)36 (6)Liraglutide3.0 mg qwkPlaceboScintigraphy16 weeksJensterle 2023, SloveniaRCTObesity200%35 (3)37 (3)Semaglutide1.0 mg qwkPlaceboScintigraphy13 weeks*Placebo group used as baseline; no other pretreatment value for gastric emptying reportedbid = twice-daily dosing; BMI = body mass index; GLP-1 RA = glucagon-like peptide-1 receptor agonist; N/A = not applicable; NR = not reported; PC = prospective cohort study; qd = once-daily dosing; qwk = once-weekly dosing; RCT = randomized controlled trial; SD = standard deviation; T1DM = type 1 diabetes mellitus; T2DM = type 2 diabetes mellitus

GLP-1 RA = glucagon-like peptide-1 receptor agonist; PRISMA = Preferred Reporting Items for Systematic Review and Meta-analyses

Demographics and characteristics of studies

*Placebo group used as baseline; no other pretreatment value for gastric emptying reported

bid = twice-daily dosing; BMI = body mass index; GLP-1 RA = glucagon-like peptide-1 receptor agonist; N/A = not applicable; NR = not reported; PC = prospective cohort study; qd = once-daily dosing; qwk = once-weekly dosing; RCT = randomized controlled trial; SD = standard deviation; T1DM = type 1 diabetes mellitus; T2DM = type 2 diabetes mellitus

Five of these studies21–25 reported two sets of results for different GLP-1 RAs or their durations. The study by Quast et al.21 presented two sets of results for two different GLP-1 RAs. As these results are derived from entirely different patient groups with no common control group, we treated them as two independent samples,19 and we distinguished them in the meta-analysis as Quast_1 and Quast_2. In addition, four studies22–25 reported two sets of dependent results for different dosages or durations of treatment, leading to within-study heterogeneity, which was addressed with a three-level meta-analysis.

The primary indication for GLP-1 RA use was type 2 diabetes mellitus in six of the studies,21,22,26–28,30 obesity in three of the studies,24,25,29 and type 1 diabetes mellitus in one study.23 Liraglutide was most frequently studied (n = 5),21,23–25,30 followed by lixisenatide (n = 3),21,27,28 exenatide (n = 2),22,28 and semaglutide (n = 1).29 Table 2 summarizes the gastric emptying T½ findings of included studies. The total sample size was 300 patients, with a mean age of 49 (11) yr and a BMI of 34 (4) kg·m−2, and 172 of them (58%) were male.Table 2Summary of gastric emptying half-time before and after glucagon-like peptide-1 receptor agonist useStudyGLP-1 RADuration of treatmentOutcome measureBaseline or control gastric emptying T½ (min)Posttreatment or treatment group gastric emptying T½ (min)Summary of resultsLinnebjerg 2008Exenatide (5 µg and 10 µg)5 daysMean (90% CI)60 (50 to 70)5 µg:111 (94 to 132)10 µg:169 (143 to 201)GLP-1 RA slowed GE of both solid and liquid meal components (P < 0.01)Lorenz 2013Lixisenatide4 weeksMean (SEM)245 (25)457 (68)GE increased substantially from baseline with lixisenatide 20 µg qd, but not with placeboBeti 2018Exenatide4 daysMean (SD)76 (24)165 (52)Gastric T½ prolonged in nearly all patients without gastroparesis at initiation with GLP-1 RA; 3 days of treatment corresponded to a 2.2-fold increase in T½Meier 2019Lixisenatide4 weeksMean (SEM)88 (12)168 (20)GLP-1 RA use prolonged GE after breakfast. Increase from 90 min at baseline to 175 min (P < 0.001)Nagai 2014Liraglutide7 daysMean (SD)107 (12)118 (17)Two patterns of response. GE delayed (delayers, n = 7); (nondelayers, n = 9). Delayers: mean increase of GE time: 31 (4) min vs baseline (P < 0.01); nondelayers: 2 (3) min vs baseline (P = 0.6).Dejgaard 2016Liraglutide3, 24 weeksMean (95% CI)95 (82 to 108)Week 3:103 (90 to 116)Week 24:84 (71 to 97)GE delayed after 3 weeks but no difference after 24 weeks of treatmentHalawi 2017Liraglutide5, 16 weeksMedian [IQR]120 [97–145]Week 5:180 [162–295]Week 16:142 [120–177]Significant effect on gastric emptying T½ of solids at both 5 and 16 weeks vs placeboQuast 2021Liraglutide.Lixisenatide10 weeksMean (SEM)Liraglutide: 98 (9)Lixisenatide88 (8)Liraglutide:122 (13)Lixisenatide:136 (16)Gastric emptying T½ prolonged by 52 min with lixisenatide and by 25 min with liraglutide (P < 0.025)Maselli 2022Liraglutide5, 16 weeksMedian [IQR]117 [98–140]Week 5:192 [137–241]Week 16:154 [120–178]At both 5 and 16 weeks, significant increase in GE vs baselineJensterle 2023Semaglutide13 weeksMedian [IQR]128 [109–142]171 [154–188]Gastric emptying T½ was longer for semaglutide vs placebo (171 vs 118 min; P < 0.001)CI = confidence interval; GE = gastric emptying; GLP-1 RA = glucagon-like peptide-1 receptor agonist; IQR = interquartile range; od = once-daily dosing; SEM = standard error of the mean; SD = standard deviation; T½ = half-time

Summary of gastric emptying half-time before and after glucagon-like peptide-1 receptor agonist use

CI = confidence interval; GE = gastric emptying; GLP-1 RA = glucagon-like peptide-1 receptor agonist; IQR = interquartile range; od = once-daily dosing; SEM = standard error of the mean; SD = standard deviation; T½ = half-time

Five studies used the stable isotope breath test,21,26–28,30 four studies used scintigraphy,22,24,25,29 and the remaining study used acetaminophen absorption.23 Three studies had a duration of less than 7 days.22,26,30 One study reported results at 21 days,23 and the remaining studies ranged from 28 to 168 days.21,23–25,27–29

Overall, meta-analysis showed a significant increase in gastric emptying T½ by a mean of 74 min (95% CI, 46 to 101) in those patients treated with GLP-1 agonists (SMD, 2.38; 95% CI, 1.05 to 3.71; P < 0.001; 95% prediction interval [PrI], −2.27 to 7.03) (Fig. 2 and Table 3).Fig. 2Three-level meta-analysis including all comparisonsCI = confidence interval; df = degrees of freedom; GLP-1 = glucagon-like peptide-1; Q = Cochrane Q statisticTable 3Estimated effect size on gastric emptying half-time classified by type of glucagon-like peptide-1 receptor agonist and duration of therapyType of GLP-1 RA and duration of therapyNumber of patientsNumber of studiesEffect size (95% CI)Converted effect size to mean difference (95% CI) (min)aAll studies300112.45 (1.54 to 3.35)74 (46 to 101)Short-actingb11153.86 (2.37 to 5.35)116 (71 to 161)Long-actingc18961.31 (0.85 to 1.77)39 (26 to 53)Duration of therapy (weeks) 1–310342.24 (− 0.88 to 5.37)67 (− 26 to 161) 4–1017762.72 (1.15 to 4.35)82 (35 to 131) > 102011.61 (0.90 to 2.33)48 (27 to 70)aDue to high heterogeneity in standard deviations, we followed Cochrane Handbook guidance (Section 6.5.1.9) and used a conservative, representative standard deviation of 30 for back-conversion to the original scalebExenatide, lixisenatidecLiraglutide, semaglutideCI = confidence interval; GLP-1 RA = glucagon-like peptide-1 receptor agonist

Three-level meta-analysis including all comparisons

CI = confidence interval; df = degrees of freedom; GLP-1 = glucagon-like peptide-1; Q = Cochrane Q statistic

Estimated effect size on gastric emptying half-time classified by type of glucagon-like peptide-1 receptor agonist and duration of therapy

aDue to high heterogeneity in standard deviations, we followed Cochrane Handbook guidance (Section 6.5.1.9) and used a conservative, representative standard deviation of 30 for back-conversion to the original scale

CI = confidence interval; GLP-1 RA = glucagon-like peptide-1 receptor agonist

Regarding heterogeneity as evaluated by I2, the presence of four studies that reported two sets of related data (different dosing regimens or duration of treatment) resulted in high heterogeneity (97.7%).22–25 We performed a sensitivity analysis excluding four within-study trials from studies with longer durations or higher dosages. This analysis similarly found a significant increase in gastric emptying T½ in those patients treated with GLP-1 RA (SMD, 2.45; 95% CI, 1.54 to 3.34; P < 0.001; 95% PrI, −0.58 to 5.48), with a heterogeneity of 95% (Fig. 3). We performed meta-regression for short-/long-acting agents, study duration, publication year, and sample size, which reduced heterogeneity further to I2 = 66%.Fig. 3Random-effect meta-analysis excluding within-study trialsGLP-1 = glucagon-like peptide-1; CI = confidence interval; RE = random effects; Q = Cochrane Q statistic; df = degrees of freedom

Random-effect meta-analysis excluding within-study trials

GLP-1 = glucagon-like peptide-1; CI = confidence interval; RE = random effects; Q = Cochrane Q statistic; df = degrees of freedom

We observed a greater prolongation of mean gastric emptying T½ for short-acting agents, such as exenatide and lixisenatide, by 116 min (95% CI, 71 to 161) (SMD, 3.86; 95% CI, 2.37 to 5.35; 95% PrI, −0.01 to 7.73), compared with long-acting agents, such as liraglutide and semaglutide, which only prolonged gastric emptying T½ by 39 min (95% CI, 25 to 53) (SMD, 1.31; 95% CI, 0.85 to 1.77; 95% PrI, 0.41 to 2.21).

In addition, we observed a greater effect in the early phases of therapy (< 10 weeks). Following 4–10 weeks of treatment with a GLP-1 RA, mean gastric emptying T½ was prolonged by 82 min (95% CI, 35 to 131) (SMD, 2.75; 95% CI, 1.15 to 4.35). This is in comparison to a mean prolongation of 48 min (95% CI, 27 to 70) (SMD, 1.61; 95% CI, 0.90 to 2.33) with treatment > 10 weeks (Table 3).

The study sample size did not impact the effect size (0.02; 95% CI, −0.02 to 0.06; P = 0.0.28; 95% PrI, −1.27 to 1.31). More recent publication years correlated with a smaller effect size (− 0.19; 95% CI, −0.42 to 0.03; P = 0.08; 95% PrI, −1.50 to 1.11). To investigate this effect in more detail, a cumulative forest plot of effect size over time is provided (Fig. 4).Fig. 4Cumulative effect meta-analysis in chronological order by publication yearCI = confidence interval; GLP-1 = glucagon-like peptide-1

Cumulative effect meta-analysis in chronological order by publication year

CI = confidence interval; GLP-1 = glucagon-like peptide-1

The overall risk of bias was low, as most studies were RCTs with good study design. Eight RCTs (assessed with RoB2) had a low risk of bias across all domains, particularly in areas of outcome data completeness, and selective reporting. Nevertheless, one study reported some concerns in the randomization process, where one treatment group had a longer duration of diabetes compared with the control group.23 Three studies presented some concerns related to lack of blinding (ESM eFig. 1).22,24,29

We assessed the two nonrandomized studies with the ROBINS-I tool (ESM eFigs. 1 and 2). One study maintained a low risk of bias across all domains.22 One study had moderate concerns with a low participation rate, no confounder adjustment, and moderate risks in selection of participants and reported results, as well as deviations from intended interventions (ESM eFig. 1).26

The funnel plot (ESM eFig. 3) and Egger’s test (P < 0.001) suggested the presence of publication bias, mainly due to low sample size and high variation. Larger effect sizes were associated with increased variance.

Following an evaluation using the GRADE system, the certainty of the evidence was rated as “very low” for all studies, as well as for the subgroups of short-acting and long-acting GLP-1 RAs, owing to concerns of publication bias; substantial heterogeneity across studies; and population, intervention, control, and outcome (PICO) indirectness, despite most included studies being RCTs. Electronic Supplementary Material eTable 2 provides the full assessment. Electronic Supplementary Material eFig. 4 presents the evaluation of the possible effect of study quality on effect size and ESM eAppendix the interaction analysis.

Seven studies comprising 196 patients aged 51 (11) yr with a BMI of 32 (5) kg·m−2 assessed the impact of GLP-1 RA on gastric emptying T½ when prescribed for diabetes mellitus.21–23,26–28,30 Four studies used a short-acting agent (two used exenatide and two lixisenatide),22,26–28 and three used a long-acting agent (liraglutide).21,23,30 A 13C-octanoic acid breath test was used to assess gastric emptying in five studies,21,26–28,30 while scintigraphy22 and acetaminophen absorption23 were used by one study each.

Of studies on a short-acting agent, gastric emptying T½ was described as increased from baseline in all studies, with percentage increase ranging from 9% to 182%. Linnebjerg et al.’s study was the only one using a short-acting agent that assessed T½ by scintigraphy and found that exenatide slowed both solid and liquid meal components.22 Additionally, a higher dose of exenatide (10 µg) resulted in a significantly longer gastric emptying T½ than a 5 µg dose.

Of the studies on a long-acting agent, liraglutide resulted in a significant prolongation in gastric emptying T½ at 3 and 10 weeks of treatment, but there was no significant delay after 24 weeks of treatment.21,23–25,30 Nagai et al. reported two patterns of responders after 7 days of liraglutide administration, with seven patients showing a significant delay in T½ and nine patients showing almost no delay despite similar baseline gastric emptying T½.31

Three studies comprising 98 patients aged 41 (12) yr with a BMI of 36 (5) kg·m−2 assessed the impact of GLP-1 RA on gastric emptying T½ when prescribed for obesity.24,25,29 All were RCTs and assessed gastric emptying by scintigraphy. Two studies used liraglutide,24,25 and one used semaglutide.29 Gastric emptying T½ was assessed after extended GLP-1 RA use (12, 13, and 16 weeks). Two studies additionally assessed gastric emptying T½ after 5 weeks of use.24,25

All studies reported a significantly prolonged gastric emptying T½ after GLP-1 RA administration regardless of the assessment method, agent, or duration of study. Two studies also reported evidence of a progressive decrease in impact on gastric emptying with liraglutide over time (median [IQR] gastric emptyingT½, 70 [32–151] min at 5 weeks vs 31 [12] min at 16 weeks).24,25 Nevertheless, gastric emptying remained significantly delayed compared with baseline at both 5 and 16 weeks.

Discussion

The results of this systematic review and meta-analysis suggest there is a significant delay in gastric emptying with GLP-1 RAs, with a mean prolongation in T½ of 74 min and an SMD > 2, suggesting a large effect size. Our analysis also suggests more pronounced effects with short-acting agents (such as exenatide and lixisenatide) and in earlier treatment phases (< 10 weeks). This delay in gastric emptying aligns with the known effects of GLP-1 RAs on reducing postprandial glucose levels by slowing nutrient absorption.31 To add a clinical context, the baseline gastric emptying T½ in the individual studies varied from 1 hr to 4 hr depending on the type of meal evaluated, and the increase in gastric emptying T½ varied between 10% and 100% posttreatment (Table 2). Our findings are consistent with a recent review by Oprea et al. and further quantify the prolongation of gastric emptying associated with GLP-1 RAs.32

Our study has several limitations. First, the variation in study designs, patient populations, and agents studied contribute to the high heterogeneity of the results. Second, the diversity in the methods of gastric emptying assessment introduces methodologic variation that may affect result comparability. For example, breath tests and scintigraphy may detect more pronounced delays in gastric emptying compared with acetaminophen absorption testing, which is influenced by intestinal absorption and primarily reflects the liquid phase of gastric emptying. Furthermore, the small number of studies reporting each assessment technique limited the scope of subgroup analyses. Most studies had small sample sizes and relatively short treatment durations, with few extending beyond 6 months. This limits our ability to comment on long-term effects of GLP-1 RAs on gastric emptying and the clinical significance of potential tachyphylaxis, and may limit the generalizability of our findings to the general population.

A further limitation is the fact that we exclusively studied gastric emptying T½ as a single parameter. This may not fully characterize the complex dynamics of gastric emptying, as the effect of some of these drugs may be more pronounced on the early stages of gastric emptying without a similar effect on total gastric emptying time. Nevertheless, we took this approach, as gastric emptying T½ was the most commonly studied metric and is a standardized parameter to evaluate gastric emptying time and compare different drugs and interventions. Additionally, the nature of the data available was unknown to us prior to the literature search. Thus, the initial analytical plan was simplified, and many details of the plan were decided post hoc, once the data became available. Finally, the baseline incidence of pulmonary aspiration in a patient who is healthy, is fasting, and is undergoing elective surgery is low (approximately 1:3,000 patients),33 and while many case reports have surfaced in the last 2 years, the absolute risk of aspiration in patients on GLP-1 RAs is still unknown. In this context, evaluating the effect of these drugs on gastric emptying time could be an informative surrogate outcome. Studies with standardized, long-term assessments of both gastric emptying phases (solid and liquid) and clinical outcomes, supported by larger sample sizes, would be beneficial to refining our understanding of the effects of GLP-1 RAs on gastric motility and the risk of pulmonary aspiration. Although all studies identified reported a significant increase for GLP-1 RAs, the certainty regarding the exact magnitude of the effect was “very low” following a GRADE assessment (ESM eTable 2).

In conclusion, in this systematic review and meta-analysis, we found that GLP-1 RAs significantly prolonged gastric emptying time by a mean of 74 min and with a large SMD (> 2), which could have implications for perioperative care. There was a trend towards more a pronounced effect with short-acting (vs long-acting) drugs and in the early treatment phases (< 10 weeks).