Comparing the Treatment and Side Effects of Existing Bariatric Surgery Procedures: An Observational Study

Bariatric surgery is an established treatment for obesity and its associated comorbidities, including diabetes, hypertension, sleep apnea, and hypercholesterolemia. Despite the widespread adoption of various bariatric procedures, rigorous causal comparisons of their differential effects on treatment outcomes and adverse events remain scarce. This large-scale observational study aimed to rigorously compare the effects of commonly performed bariatric surgery procedures on both weight loss (effevtiveness) and the risk of postoperative complications. This study utilized data from the American College of Surgeons Metabolic and Bariatric Surgery Accreditation and Quality Improvement Program (MB-SAQIP) database, encompassing 729,482 cases from 2015 to 2020. With Sleeve Gastrectomy serving as the reference procedure, we assessed the effect of alternative procedures on changes in body mass index (BMI) and the risk of reoperation, readmission, and subsequent interventions. State-of-the-art machine learning-based causal inference techniques, including Causal Forest, Dragonnet, and Double Machine Learning, were employed to conduct robust causal comparisons. Biliopancreatic Diversion with Duodenal Switch (BPD/DS) demonstrated superior BMI reduction compared with Sleeve Gastrectomy. Roux-en-Y Gastric Bypass (RYGB), Adjustable Gastric Band (AGB, or Band), and Single Anastomosis Duodeno-Ileal Bypass with Sleeve Gastrectomy (SADI-S) were associated with less pronounced BMI decreases relative to Sleeve Gastrectomy. The risk of complications was similar or higher for all other surgical procedures compared with Sleeve Gastrectomy. Importantly, these represent causal effect estimates rather than mere associations, providing clinically actionable evidence for treatment selection. Detailed effect estimates and risk ratios, along with their confidence intervals, are presented in the full text. All our implementations are available at GitHub. Our causal estimates–derived from state-of-the-art machine learning methods applied to the largest bariatric surgery registry–provide the first rigorous quantitative evidence supporting current clinical practice guidelines, issued by the American Society for Metabolic and Bariatric Surgery (ASMBS), and enable evidence-based surgical decision-making. What are the causal effects of the five widely adopted bariatric surgery procedures on weight loss efficacy and postoperative complication risks? Our causal analysis reveals that Biliopancreatic Diversion with Duodenal Switch (BPD/DS) achieves significantly greater BMI reduction compared with the most widely conducted Sleeve Gastrectomy, but at the cost of substantially elevated complication risks. Our causal analysis results of all five bariatric surgery procedures align with mechanistic understanding and provide quantitative causal estimates rather than associations. This represents the first large-scale and comprehensive causal analysis comparing weight loss and adverse event risks across the five most important bariatric surgery procedures, providing rigorous evidence to inform surgical decision-making.