Palliative treatment for gastric outlet obstruction: A network meta-analysis.

Abstract

36 Background: The optimal palliative treatment for gastric outlet obstruction (GOO) remains inconclusive between gastrojejunostomy (G), endoscopic ultrasound-guided gastroenterostomy (E), stomach partitioning gastrojejunostomy (P), endoscopic stenting (S). This study was part of a comprehensive systematic review investigating the outcomes of the aforementioned treatments for malignant GOO. Methods: We conducted a systematic screening randomized controlled trials (RCTs) and cohort studies that compared at least two palliative procedures for GOO from Pubmed, Embase, Cochrane, Web of Science, Scopus, Clinicaltrial, WHO International Clinical Trials Registry Platform. We included full-text studies that reported at least clinical outcomes (clinical success rate, complication rate, 30-day mortality rate, and reintervention rate). We conducted this network meta-analysis using the frequentist approach, inverse variance model with a naïve combination of the treatment effects from RCTs and non-RCTs. We used P-score for treatment ranking. Certainty of evidence was evaluated following CINeMA approach. Results: This study included four RCTs and four prospective and 32 retrospective cohorts with 3417 patients. The pooled overall rates of clinical success, complication, 30-day mortality, and reintervention were 88.9% (95%CI 85.6-91.6), 20.7% (95%CI 17.2-24.7), 5.4% (95%CI 3.2-8.9), and 13.9% (95%CI 10.7-17.9), respectively. P was ranked the safest for reintervention rate (P-score: 0.90) due to obstruction and complication post-procedure (P-score: 0.88 and 0.81, respectively). E was ranked the safest for the 30-day mortality rate (P-score: 0.82). Cluster rank combined the P-score for 30-day mortality and reintervention or reintervention rate due to obstruction showed the benefit of P and E versus G and S (cophenetic correlation coefficient - c: 0.94 and 0.94, respectively). Cluster rank combined with the P-score for 30-day mortality and reintervention due to complication showed the benefit of P (c:0.99). The overall certainty of evidence was low to very low. Conclusions: P and E are recommended for malignant GOO, and P should be the first choice in centers with limited resources or cases of unfeasible or unsuccessful E.