- Design
- Bayesian network meta-analysis of 7 RCTs and 8 cohort studies
- Population
- 827 adults with malignant gastric outlet obstruction
- Primary outcome
- Clinical success, reintervention and recovery
- Effect
- EUS-GE vs surgical GJ clinical success RR 1.38 (95% CrI 1.15–1.88); stent vs EUS-GE reintervention RR 7.56 (2.02–38.14)
This Bayesian network meta-analysis pooled 15 studies — seven randomised trials and eight comparative cohorts, 827 patients — comparing surgical gastrojejunostomy, endoscopic duodenal stenting and EUS-guided gastroenterostomy for malignant gastric outlet obstruction.
EUS-guided gastroenterostomy had higher clinical success than surgical gastrojejunostomy (RR 1.38, 95% credible interval 1.15 to 1.88), and stenting was also better than surgery on that measure (RR 1.18). Stenting needed reintervention far more often than either surgery (RR 3.61) or EUS-guided gastroenterostomy (RR 7.56). Both endoscopic options shortened time to eating and hospital stay. Adverse events, 30-day mortality and overall survival did not differ.
For a surgeon, the practical point is that EUS-guided gastroenterostomy is now a reasonable option to discuss in centres with the expertise, particularly for patients expected to live long enough for stent failure to matter. Surgical gastrojejunostomy remains established where endoscopic expertise is unavailable, which is still true in many Indian centres.
- Discuss EUS-guided gastroenterostomy with gastroenterology for malignant outlet obstruction where local expertise exists.
- Reserve duodenal stenting for patients with short expected survival, given its higher reintervention rate.
- Keep surgical gastrojejunostomy as the option where advanced endoscopic expertise is not available.
- Agree the choice in a multidisciplinary meeting, weighing prognosis and fitness.
Why it matters
It shifts the default palliative referral from the operating theatre towards the endoscopy suite where expertise allows.
Don't overread it
Over half the studies were cohorts and numbers were small; direct randomised comparisons are still needed.
The statistics, in plain English
A credible interval is the Bayesian version of a confidence interval. Here the reintervention estimates have very wide intervals (for example 2.02 to 38.14), meaning the direction is clear but the size is not. Mixing cohort studies with trials adds risk of bias.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for general surgery, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free