GLP-1 agonists and the adolescent stomach, COX-2 inhibitors measured properly, cryoanalgesia levelled with epidural, and root preservation in type A dissection

A genuinely good day: four items, every one published on 8 or 9 September, all with deposited abstracts read in full. Nothing older mixed in today — there was no need.

1. GLP-1 receptor agonists in adolescents — 80% had solids after a twelve-hour fast

Journal · Published: British Journal of Anaesthesia, 8 September 2026 — published yesterday O’Brien, Gallop, Washburn, Dumas, Como, Kilungya, Willi, Lin

Single-institution prospective cohort, June 2023 to November 2024, in adolescents aged 10–18. GLP-1 receptor agonists are approved from age 10 for type 2 diabetes and 12 for obesity, so this population is growing quickly.

Three groups, 67 patients analysed: 20 on a GLP-1 receptor agonist, 27 at risk of delayed gastric emptying but not on one, and 20 healthy controls. All fasted to standard guidance — 8 hours for solids, 1 hour for clear fluids — and gastric contents were assessed by ultrasound, qualitatively on a 3-point scale and quantitatively. The primary outcome was residual gastric contents, defined as solids present or gastric fluid volume ≥1.5 ml/kg.

The actual fasts were far longer than required: median 13 hours for solids (IQR 12–14) and 12 hours for clear fluids (IQR 3.5–13).

The result:

• GLP-1 group: 16 of 20 (80%) had solids on gastric ultrasound
• At-risk group not on GLP-1 agonists: 17 of 27 (63%)
• Healthy controls: 1 of 20 (5%)
• Unadjusted P < 0.001; P < 0.02 after propensity-score adjustment for age, BMI, sex, race and NPO time

The authors’ own summary: “80% of adolescents on GLP1-RAs had solids on gastric ultrasound despite having fasted for over 12 h.”

Interpretation. This is the most directly actionable finding this archive has carried, and it is worth being precise about why.

Doubling the fast did not empty the stomach. These children fasted a median of 13 hours — five hours beyond the guideline — and four in five still had solid contents. Whatever the right answer is for this population, “fast them longer” is not it. That matters because prolonged fasting is the intervention most units reach for first, and it is the one this data specifically rules out.

The 63% in the at-risk comparison group is the finding people will miss. Adolescents with risk factors for delayed gastric emptying but not on a GLP-1 agonist were also overwhelmingly likely to have solids present. The gap between them and the GLP-1 group (80% vs 63%) is much smaller than the gap between either and healthy controls (5%). Read strictly, the drug is not creating a unique hazard so much as adding to a hazard that was already there and largely unrecognised — which is an argument for gastric ultrasound in the at-risk adolescent generally, not only in the ones on semaglutide.

The honest limits: 67 patients at one centre, with 20 in the index group, and the outcome is a sonographic surrogate, not aspiration. Nobody aspirated in this study because it was not powered to see that, and the leap from “solids visible” to “will aspirate” is exactly the leap the adult GLP-1 literature has been arguing about for three years. What this adds is that the paediatric and adolescent version of the question has now been asked, and the answer looks worse than in adults, not better.

Practically: for an adolescent on a GLP-1 agonist, this supports point-of-care gastric ultrasound rather than a longer fast, and treating a positive scan as a full-stomach airway plan.

Read the paper · PMID 42711215

2. COX-2 inhibitors — the first meta-analysis to ask whether the pain interferes

Journal · Published: Anesthesiology, 9 September 2026 — published today Kahloon, Le, Tierney, Fergusson, Lalu, McIsaac, Turgeon, Patel, Musa, Verret

MEDLINE, Embase and Cochrane CENTRAL, systemic perioperative selective COX-2 inhibitors versus placebo, opioids or usual care in adults. 9,071 citations screened, 38 RCTs, 5,424 participants.

The methodological point is the primary outcome. Rather than a numeric rating score, it is acute pain measured on a validated multidimensional instrument capturing interference with daily living — Brief Pain Inventory or equivalent, 0–10, with a prespecified minimal clinically important difference of 1.0.

Results:

• Acute pain-related interference: BPI mean difference −1.1 (95% CI −1.4 to −0.8), 7 trials, moderate certainty — the point estimate exceeds the MCID and the whole confidence interval sits below zero
• Chronic pain incidence: OR 0.44 (95% CI 0.21 to 0.93), 5 trials, low certainty
• Chronic pain intensity: MD −1.0 (95% CI −2.7 to 0.7), 1 trial, very low certainty — null
• Quality of Recovery-9: MD 0.84 (95% CI 0.33 to 1.35) on an 18-point scale, low certainty
• Intraoperative blood loss: −22 ml (95% CI −40 to −4), low certainty
• No difference in acute renal failure, gastrointestinal bleeding, impaired bone healing, myocardial infarction, stroke or death

Interpretation. Two things here, and the second is bigger than the first.

The first is the result itself, which is reassuring but not surprising: COX-2 inhibitors work, the effect on function clears the clinically important threshold, and the feared harms did not appear. The −22 ml blood loss figure is worth quoting the next time someone withholds a coxib from a surgical patient on bleeding grounds — it is a statistically significant difference of no clinical consequence whatsoever, in the favourable direction, and it is a useful antidote to the reflex that groups all NSAIDs together.

The second is the choice of outcome, and it is the reason this paper is worth reading rather than filing. Perioperative analgesia trials have measured pain scores for forty years, and pain scores are a poor guide to whether the patient can get out of bed, sleep, or breathe deeply. A mean difference of −1.1 on interference with daily functioning is a claim about the patient’s day, not about a number they said out loud. Note also that the interference outcome rests on 7 trials of the 38 — most of the literature still does not measure this — and that the certainty grading is moderate for the primary and low or very low for everything else.

Set it beside the ketamine analgosedation trial sent yesterday, which reported a Bayesian probability of benefit on opioid sparing. Both are attempts to answer “does this matter to the patient” with something better than a 0–10 score, and both are more informative for it.

Read the meta-analysis · PMID 42708838

3. Cryoanalgesia, ESP block and epidural after minimally invasive lung resection — a three-way draw

Journal · Published: Updates in Surgery, 8 September 2026 — published yesterday Mammana, Comacchio, Sella, Pagliarini, Shamshoum, Falcioni, Zanardo, Silvestrin, Brezzi, Pittarello, Nicotra, Dell’Amore (Padua)

Randomised controlled trial, 25 patients per group, in minimally invasive anatomic lung resection: cryoanalgesia versus erector spinae plane block versus epidural analgesia. Static and dynamic pain scores to two months; primary outcome static NRS at 24 hours.

• Primary outcome null: static NRS at 24 h 1.92 (±2.12) cryoanalgesia, 2.63 (±2.14) ESPB, 2.45 (±1.95) epidural, P = 0.494
• Epidural was associated with lower intraoperative remifentanil and a trend toward better patient satisfaction at 2 months
• Pain scores at all other time points and postoperative opioid consumption were comparable
• Authors’ conclusion: no significant differences over the first 48 hours or at 2 months; “further studies are needed in order to understand how to improve the efficacy of cryoanalgesia.”

Interpretation. The interesting part is the absolute numbers, not the P value. Every group sat below NRS 3 at 24 hours after anatomic lung resection. When all three arms land in well-controlled territory, a null primary outcome tells you the operation has changed more than the analgesia has — minimally invasive anatomic resection is simply not the pain problem that thoracotomy was, and the three-way tie is partly a floor effect.

That reframes the choice. If analgesic efficacy is equivalent, the decision falls to everything else: an ESP block is quick, superficial and compatible with anticoagulation in a way an epidural is not; cryoanalgesia is performed by the surgeon under direct vision and needs no catheter, but carries the intercostal numbness and the neuropathic-pain question that this trial’s two-month follow-up was too small to settle. The epidural’s advantage here was intraoperative — lower remifentanil — which is a real benefit and also the one most easily replaced by other means.

Read against the BJA paravertebral-versus-ESP noninferiority trial in cardiac surgery, the pattern is becoming familiar: fascial plane blocks keep matching the techniques they were meant to be a compromise for. The honest summary across both is that the deeper, riskier block is no longer buying much, in populations where the surgery itself has got less painful.

Limits worth stating: 25 per group is small, the primary outcome is a single static score at one time point, and a trial this size cannot exclude a moderate difference. It is hypothesis-shaping, not practice-changing.

Read the trial · PMID 42709387

4. Aortic root preservation in acute type A dissection — 1,333 patients, ten-year data

Journal · Published: Interdisciplinary CardioVascular and Thoracic Surgery, 8 September 2026 — published yesterday Zuo, Li, Zhai, Dong, Lai, Wang (Zhongshan Hospital, Fudan)

Single-centre series, 2005–2024, of surgical repair for acute type A aortic dissection with root preservation, stratified by preoperative aortic regurgitation. Root preservation was the default unless limited by severely damaged root intima, significant AR with cusp lesions, dilated sinus, or connective tissue disorder.

1,333 patients — 369 with moderate or severe AR, 964 with less than moderate.

• 30-day mortality 11.3% overall — 15.2% in the AR group versus 9.9% (P = 0.006) — but preoperative AR was not associated with early mortality after multivariable adjustment
• 10-year survival comparable: 93.6% versus 92.3%
• Freedom from recurrent AR at 10 years: 88% (AR group) versus 97.1% (P < 0.001)
• Strongest predictors of recurrence: intraoperative residual mild AR (HR 4.08, P = 0.001) and connective tissue disorders (HR 4.36, P = 0.018)
• Without intraoperative residual mild AR, preoperative moderate/severe AR did not increase the risk of late recurrence

Interpretation. The finding that matters is the one about the operating room, not the one about the patient.

Preoperative AR is not the risk factor. Residual AR at the end of the case is — a hazard ratio of 4.08 for what is, by definition, mild regurgitation on the completion assessment. And the corollary is stated explicitly: with no residual mild AR, a patient who arrived with moderate or severe AR had no excess late recurrence. The preoperative echo does not determine the durability of a preserved root; the post-bypass echo does.

For anyone doing transoesophageal echo in these cases, that converts “trivial-to-mild residual AR, acceptable” into a decision point. It is a hard call in a bleeding, often unstable acute dissection at the end of a long case — and this series says that accepting it multiplies the risk of late recurrence roughly fourfold.

The cautions are the usual ones for a single-centre observational series, and one specific one: this is a selected population. Roots were preserved only when they were preservable, so the excellent 10-year survival is partly a statement about who was chosen. The paper does not answer whether preservation beats replacement — it answers what happens to the roots that were preserved. It sits naturally beside the JTCVS type A dissection series sent on 31 August on prior cardiac surgery and dissection outcomes.

Also note 30-day mortality of 11.3% in a modern high-volume series. Acute type A dissection remains what it has always been.

Read the paper · PMID 42711262

---

Notes for the next run

• Seen and not pursued. Regional Anesthesia and Pain Medicine published a full themed issue on implanted intrathecal drug delivery systems on 8 September — fourteen papers. Most is chronic pain rather than primary scope, but one is squarely on topic and worth a slot on a quiet day: Jenkinson et al., “Implanted intrathecal drug delivery systems: a narrative review for the perioperative, obstetric, and intensive care anesthesiologist” (DOI 10.1136/rapm-2025-107458, PMID 42710951) — covers neuraxial access precautions in patients with an IDDS, intrathecal baclofen withdrawal, and MRI behaviour of the devices.
• ITACS appeared in the BMJ sweep dated 7 September and was correctly skipped — it was sent on 25 August. Europe PMC’s first-publication date reflects the BMJ print version, not the original release. This is the aggregator date problem in a new place; recorded in the source notes.
• Also seen, out of primary scope: an Annals of Surgery global analysis of hospital-level determinants of mortality after pancreas surgery, a JAMA research summary on gastric residual volume in critically ill children, and an EJA pair on single- versus multi-orifice catheters for dural puncture epidural labour analgesia.