Gastric partitioning compared to conventional gastrojejunostomy as palliative surgeries in patients with gastric outlet obstruction: a pairwise and individual patient data meta-analysis

BACKGROUND: Gastric outlet obstruction (GOO) complicates unresectable gastric and pancreatic cancers. Conventional gastrojejunostomy (CGJ) is standard but frequently leads to delayed gastric emptying. Stomach-partitioning gastrojejunostomy (SPGJ) mitigates this problem and improves outcomes. METHODS: We conducted a meta-analysis of SPGJ versus CGJ for GOO, searching databases through 25 November 2025. Outcomes were delayed gastric emptying (DGE), major complications, reintervention, 30-day mortality, operative time, Gastric Outlet Obstruction Scoring System (GOOS) scores, length of stay, chemotherapy adherence, and survival. Continuous variables were pooled as mean differences (MD) with 95% CIs; dichotomous variables as relative risks (RR) with 95% CIs. Survival was analyzed using individual patient data reconstructed from Kaplan-Meier curves. RESULTS: A total of 11 studies comprising 456 patients were included. SPGJ was associated with significantly reduced DGE (RR = 0.24, 95% CI: 0.12-0.47) and postoperative major complications (RR = 0.26, 95% CI: 0.12-0.54) compared to CGJ. No significant differences were found in the need for reintervention (RR = 0.59, 95% CI: 0.21-1.64), short-term mortality (RR = 0.99, 95% CI: 0.42-2.33), or LOS (MD = -1.47 days, 95% CI: -3.10 to 0.16). GOOS scores were comparable between groups. Overall survival was also similar between SPGJ and CGJ (HR = 1.06, 95% CI: 0.66-1.70). CONCLUSIONS: Our meta-analysis shows that SPGJ offers important clinical advantages over CGJ by significantly reducing delayed gastric emptying and postoperative major complications, while demonstrating comparable GOOS scores, length of stay, reintervention rates, and short- and long-term survival. These findings support SPGJ as a viable and potentially preferable option for managing malignant GOO, although high-quality randomized trials are still needed.