S1278 Comparative Efficacy and Safety of Gastrostomy Insertion Techniques: A Network Meta-Analysis

Introduction: Current guidelines, such as those from the 2025 American Society for Gastrointestinal Endoscopy, recommend percutaneous endoscopic gastrostomy (PEG) as the primary technique for enteral access. However, limited comparative evidence across insertion methods results in practice variability and inconsistent patient outcomes. Thus, this network meta-analysis was done to compare efficacy and safety of various gastrostomy insertion techniques. Methods: A systematic search of PubMed, Embase, Cochrane, and Scopus was done for studies comparing at least 2 gastrostomy techniques in adults undergoing first-time placement. Evaluated techniques included Pull PEG, Push PEG, Introducer PEG, percutaneous radiologic gastrostomy (PRG), ultrasound-guided (PUG), laparoscopic (LapGT), surgical (SG), and per oral image-guided (PIG). Patients with GIT malignancies or <18 years were excluded due to tumor related confounding and pediatric anatomical differences. Statistical analysis was conducted on R (v4.4.3), applying odds ratios (OR) for binary and standardized mean differences (SMD) for continuous outcomes. PEG Pull was set as the reference. Results: From the 38 studies and 38,635 patients analyzed, PIG had the highest technical success (OR = 4.14, CI: 1.76–9.74, P = 0.0012), followed by PRG (OR = 1.66, CI: 0.94–2.92, P = 0.078). However, PRG significantly increased major (OR = 1.14, CI: 1.09–1.18, P < 0.0001) and minor complications (OR = 2.42, CI: 1.63–3.39, P < 0.0001). PEG Introducer significantly reduced minor complications (OR = 0.09, CI: 0.00–0.76, P = 0.0018) and ranked best in reducing peristomal infections (OR = 0.29, CI: 0.08–0.77, P = 0.081). Conversely, PIG showed highest OR for peristomal infections (OR = 2.04, CI: 1.25–5.52, P = 0.22). SG presented the lowest risk for infections (OR = 0.46, CI: 0.14–2.29, P = 0.14), but significantly increased tube dislodgement (OR = 4.10, CI: 1.42–11.80, P = 0.0091) and procedure-related mortality (OR = 2.02, CI: 1.26-4.79, P = 0.0091). PRG ranked highest for shortest procedure duration, (SMD = -1.34, CI: -3.60-0.92, P = 0.24) while LapGT ranked lowest (SMD = 1.78 (95% CI: -0.30-3.87, P = 0.094). Netsplit analysis and net heat diagrams demonstrated minimal model inconsistency. Heterogeneity was low (I² < 35%), except for procedure duration. Conclusion: PIG demonstrated the highest technical success but with increased infection risk. PEG Introducer showed the best safety profile, while PRG is efficient but has more complications. These findings facilitate technique selection, and inform guideline refinement.