SRS394 - Defining quality in robotic pancreas surgery: predictors of textbook outcomes in a multicentre UK series

Abstract Background Textbook outcome (TO) is an objective benchmark of surgical quality. Current Brescia guidelines suggest proficiency in robotic left pancreatectomy (RLP) after 21 cases for morbidity risk, but the influence of learning phase on TO has not been validated in multicentre series. Methods All RLPs performed across six UK centres (July 2014–August 2024) were analysed. TO was defined as absence of major morbidity (complication ≥Clavien-Dindo 3a), in-hospital mortality, 90-day readmission, prolonged length of stay, and clinically relevant postoperative pancreatic fistula (CR-POPF). Predictors of TO were assessed using multivariable logistic regression. Results A total of 281 patients underwent RLP, of whom 143 (50.9%) achieved TO. Major morbidity occurred in 13.5%, CR-POPF in 20.3%, and 90-day readmission in 24.2%. There was no difference in TO between competency (n = 109) and proficiency (n = 172) phases (52.3% versus 50.0%, P = 0.801). The proficiency phase was characterised by longer operative time (315 versus 230 min, P < 0.001), lower splenic preservation (13.4% versus 24.8%, P = 0.023), and fewer resections for vascular involvement (7.0% versus 20.2%, P = 0.002). On multivariable analysis, prolonged operative duration independently reduced the likelihood of TO (OR 0.82 per additional hour, P = 0.010), with a non-linear association observed. Conclusions In this first multicentre UK series, half of patients achieved TO after RLP. Contrary to guideline assumptions, TO rates were unaffected by learning phase, likely reflecting prior laparoscopic expertise. Operative duration emerged as the strongest predictor of TO, serving as a surrogate for case complexity and highlighting the importance of patient selection and procedural standardisation as robotic pancreas surgery evolves.