health · family: success's shadow
the stomach that won't empty
The drug slows the stomach — and fasting alone does not show which patient still has food in it
Problem statement
GLP-1 receptor agonists (semaglutide and its relatives) lower blood sugar and make people feel full, and they also slow how fast the stomach empties. That slowing means a patient who has followed every fasting instruction before surgery, endoscopy or sedation may still have solid food sitting in the stomach when the anaesthetist puts them to sleep, which is the classic setup for regurgitation and pulmonary aspiration. In a prospective cross-sectional study of 124 fasted patients, increased residual gastric content was found on ultrasound in 56% of GLP-1 users versus 19% of non-users (adjusted prevalence ratio 2.48, 95% CI 1.23–4.97). The unsolved problem is that the bedside test that can show which individual patient on these drugs actually has a full stomach on the day of the procedure, gastric ultrasound, is limited (the guidance says) by institutional resources, variation between users, and credentialing requirements — so a multi-society guidance document (ASA, AGA, ASMBS, ISPCOP, SAGES) that itself says it "should be considered guidance, and not an evidence-based guideline" is what stands between a fast-growing patient population and an airway emergency.
Why this matters
Use of GLP-1 receptor agonists for diabetes and weight management is rapidly increasing in the US, and the JAMA Surgery authors expect clinicians to meet these drugs with increasing frequency around surgery — meaning that a drug class introduced for metabolic disease has become an anaesthetic-risk variable in operating theatres and endoscopy suites. The cost of not solving it is paid in three currencies: rare but devastating aspiration events; procedures cancelled on the day when there is concern about retained gastric contents; and rapid-sequence inductions or 24-hour liquid-diet preparations applied to patients who may have had no retained contents at all (in the JAMA Surgery study, increased residual gastric content was found in 35 of the 62 GLP-1 users, not in all of them). The guidance says that earlier documents from different organisations were inconsistent and left providers uncertain, and it asks individual practices to develop their own protocols.
What’s been tried and why it hasn’t worked
Three approaches are in circulation and each fails in a distinct way. (1) Hold the drug. The 2023 ASA advice was to withhold weekly agents for a week before surgery, but the multi-society document is explicit that "the duration to hold therapy is unknown"; stopping is also not free, since the guidance warns it can bring on hyperglycaemia, and, for weekly formulations with long half-lives, may not clear the effect anyway (in the JAMA Surgery study, the prevalence of increased residual content was still elevated after 7 days off the drug). (2) Change the preparation. The guidance recommends a "preoperative liquid diet for at least 24 h, as performed in patients undergoing colonoscopy and bariatric surgery" when delayed emptying is a concern — but this is borrowed from the preparation used for colonoscopy and bariatric surgery, and the guidance itself says little evidence exists to guide the best approach to managing these drugs around a procedure. (3) Measure the stomach. Point-of-care gastric ultrasound is the bedside test the guidance names for assessing aspiration risk on the day of the procedure, and the guidance concedes it "may be clinically limited based on institutional resources, inter-user variability, and credentialing requirements" — i.e. it depends on the operator, on who is credentialed, and on what the institution has. Underneath all three is an evidence problem: aspiration is rare (the JAMA Surgery authors give an incidence of 1 in 3000 to 4000 elective procedures), so their study did not assess aspiration events directly and used residual gastric content instead. The JAMA Surgery authors flag exactly this — increased residual content is a surrogate, not an aspiration event.
What would unlock progress
The bottleneck is a triage measurement, not a drug decision: something that reliably sorts "empty enough" from "not empty" at the bedside, in minutes, without a credentialed sonographer. Two directions are open. One is automated interpretation of gastric ultrasound images, so that the exam depends less on a credentialed expert; neither source reports that this has been done, and it is offered here as a proposal. The other starts from the guidance's own list of things that raise risk — the dose-escalation phase, higher dose, weekly dosing, gastrointestinal symptoms, and other conditions that delay gastric emptying. The guidance gives these as considerations for shared decision-making, not as a validated scoring rule; a stratification rule built from them and validated would let hospitals reserve scarce ultrasound for the genuinely uncertain middle.
Entry points for student teams
A team could build and evaluate an automated antral-image classifier (or a phone-based capture-and-quality-check aid) against expert gastric ultrasound reads on a public or partner image set, reporting sensitivity for solid contents rather than accuracy alone. A clinical-informatics team could derive a candidate risk-stratification rule from an existing perioperative dataset — which combinations of agent, dose interval, and symptoms predict residual content — and pre-specify how it would be validated. A human-factors team could redesign the pre-procedure instruction pathway itself: what a patient on a weekly GLP-1 is told, when, and by whom, measured by adherence and by cancellation rate. Relevant skills: signal/image processing and machine learning, clinical epidemiology, human factors, and health-service design.
Genes
Structural cousins: stuck the same way, in other fields
Sources
Kindel TL, et al., "Multi-society clinical practice guidance for the safe use of glucagon-like peptide-1 receptor agonists in the perioperative period," *Surgical Endoscopy* 2025;39(1):180–183 (published online 29 October 2024), (doi 10.1007/s00464-024-11263-2), accessed 2026-08-18; Sen S, Potnuru PP, Hernandez N, et al., "Glucagon-Like Peptide-1 Receptor Agonist Use and Residual Gastric Content Before Anesthesia," *JAMA Surgery* 2024;159(6):660–667, (doi 10.1001/jamasurg.2024.0111), accessed 2026-08-18 go to source 1 ↗ go to source 2 ↗
Working notes
Notes made while writing and checking this brief, shown as written. They use working shorthand.
Both sources were read in full text on 2026-08-18. All quoted phrases ("should be considered guidance, and not an evidence-based guideline"; "the duration to hold therapy is unknown"; "preoperative liquid diet for at least 24 h…"; "may be clinically limited based on institutional resources, inter-user variability, and credentialing requirements") are from the multi-society guidance; the 56%/19% prevalence and the 2.48 (1.23–4.97) adjusted prevalence ratio are from Sen et al. JAMA Surgery 2024. Note for verifiers: the magnitude of GLP-1-induced delay in gastric emptying at steady state is not settled in the two sources read (the guidance says little evidence exists to guide perioperative management; Sen et al. describe the handful of earlier studies of residual gastric content in GLP-1 users as having mixed results); that uncertainty is the reason this brief exists (fasting alone does not identify the at-risk individual), and this brief deliberately does not assert a settled effect size beyond the two sources read.
`failure:success-caused` is applied under the discriminating test: (1) the intervention succeeded at its stated objective — GLP-1 RAs achieve glycaemic control and weight loss at scale; (2) the harm arises through a specific named mechanism — delayed gastric emptying leaving solid residue past the fasting interval; (3) the harm is structurally coupled to the success, because the slowing of gastric emptying and the metabolic effects both come from GLP-1 receptor signalling (the guidance says GLP-1 acts on its receptor in the pancreas, brain and stomach, and in the gut increases gastric emptying time). `failure:wrong-problem` is the structural opposite and does not apply (the right problem was solved). `failure:regulatory-mismatch` was considered and rejected: no approval pathway is missing and no regulation created the perverse incentive; the gap is clinical evidence and bedside measurement. `temporal:newly-created` is applied per the failure-tag decision order step C: no specific prior technical barrier lifted (so not `tech-limitation-now-resolved`); the problem itself only emerged with mass adoption of this drug class. `constraint:data` is primary (the outcome evidence linking any strategy to aspiration does not exist and is hard to generate because the event is rare); `constraint:technical` is secondary (operator-dependent ultrasound with no scalable substitute). No existing brief covers this; the nearest neighbours are `health-home-use-device-safety` and `health-patient-designed-device-clinician-mismatch`, both unrelated in mechanism.
Source type: Self-articulated (professional societies stating the limits of their own guidance) plus primary observational evidence.
Verified at intake 2026-08-18: gate (net) + adversarial source check + contested-tag second coding. Verifier: both PMC full texts fetched 2026-08-18; all four guidance quotations, the 124/56%/19% figures and the 2.48 (1.23–4.97) aPR confirmed verbatim; Sen et al. describe a cross-sectional design with prospective enrolment.
Reconciliation 2026-10-01: Fact repair after reading the whole write-up against both of its sources, fetched 2026-10-01. The multi-society guidance (Kindel et al.) was read in full text from Europe PMC (https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11666732/fullTextXML) and from the PMC page on the Source line. For Sen et al., the PMC page returned a robot check, so the full text was read from the Internet Archive copy of the publisher page (web.archive.org/web/20260223123236/jamanetwork.com/journals/jamasurgery/fullarticle/2815663), with the abstract and citation also read from the Europe PMC record; the supplement was not read. (1) The long title said "The Drug Works Because It Slows the Stomach — and Nobody Can Tell Which Fasted Patient Still Has Food in It." Neither source says the drugs work because they slow the stomach: the guidance lists lower blood sugar and enhanced satiety as the effects, and separately says GLP-1 signals increase gastric emptying time. And Sen et al. call gastric ultrasound a validated point-of-care technique and used it to tell. The long title is now "The Drug Slows the Stomach — and Fasting Alone Does Not Show Which Patient Still Has Food in It." The display title is unchanged. The first sentence of the Problem Statement, and the matching clause in the tag reasoning above, were changed the same way. (2) "There is no practical, validated way to tell" which patient has a full stomach was too strong for the same reason; the sentence now says the bedside test that can show it, gastric ultrasound, is limited by the three things the guidance names: institutional resources, variation between users, and credentialing requirements. (3) The Source line gave the guidance as Surgical Endoscopy 2024;39(1). The Crossref and Europe PMC records give volume 39, issue 1, pages 180–183, as the January 2025 issue, published online 29 October 2024; the Source line now says that. (4) "Tens of millions of people" on these drugs, presenting for procedures "at ordinary population rates," is not in either source and was removed; the sentence now gives what Sen et al. say, that use is rapidly increasing in the US and that clinicians will meet these drugs with increasing frequency around surgery. (5) "A much larger volume of same-day cancellations" and "unnecessary" inductions and liquid diets are not in the sources. The guidance names procedure cancellation and rapid-sequence induction as the options on the day; the sentence now says that, and adds the study's own count that increased residual content was found in 35 of the 62 GLP-1 users, not in all of them. (6) "Practice varies between hospitals within the same city" is not in the sources; the sentence now gives what the guidance says, that earlier documents were inconsistent and that practices should develop their own protocols. (7) The 2023 ASA advice "was widely adopted" is not in the sources and was removed. "Destabilises glycaemic control" now follows the guidance, which warns of hyperglycaemia; the remark that a one-week hold may not clear the effect now cites the study's finding that the prevalence of increased residual content was still elevated after 7 days off the drug. (8) The liquid diet "is burdensome, and has not been shown to reliably empty the GLP-1 stomach" is not in the sources; the sentence now says the practice is borrowed from colonoscopy and bariatric surgery preparation and that the guidance says little evidence exists to guide the best approach. (9) Gastric ultrasound as "the only bedside test" and as "unavailable at the scale required" went beyond the guidance; the sentence now keeps to the guidance's three stated limits. (10) "Requires enormous cohorts" and "the available studies use residual gastric content as a surrogate" were generalisations; the sentence now gives the study's own figure for how rare aspiration is (1 in 3000 to 4000 elective procedures) and says this study did not assess aspiration events. (11) The statements that automated ultrasound reading "has already been shown" to work for cardiac and lung exams, that the gastric antrum is "a comparatively simple target," and that a risk rule "has never been built and validated for this population" are not in either source. The first is now offered plainly as a proposal; the second now starts from the risk factors the guidance lists (escalation phase, higher dose, weekly dosing, gastrointestinal symptoms, other conditions that delay emptying). (12) The note above said the size of the delay "is actively contested in the 2025–2026 literature"; both sources date from 2024, so it now says the size of the delay is not settled in the two sources read, with the guidance's statement that little evidence exists and Sen et al.'s that the few earlier studies had mixed results. Confirmed in the sources and left as written: all four quotations from the guidance, word for word; the five societies; the Sen et al. citation (JAMA Surgery 2024;159(6):660–667); 124 fasted patients, 56% against 19%, and the adjusted prevalence ratio of 2.48 (1.23–4.97); the cross-sectional design with prospective enrolment; the one-week hold for weekly agents in the 2023 ASA advice; and residual content as a surrogate for aspiration. The entry points make no factual claims that could be checked against the sources and were left as written; no public gastric ultrasound image set is named, so a team should expect to need a partner for images. Genome tags untouched. A second reader went over these repairs on the same day with both sources open (the guidance from Europe PMC and the PMC page; Sen et al. in full from the PMC page, which served the article rather than a robot check): every repair stood on its source, and six were tightened to say no more than the source does (the ultrasound sentence in the Problem Statement, "in the US" on the growth of use, the 35-of-62 count in place of a derived 27, the prevalence wording for the 7-day finding, the tag reasoning, and the verifier note).