Towards standardized assessment of surgical difficulty in robotic hepatobiliary surgery: a comparative validation study

The advent of robotic platforms in hepatobiliary surgery has expanded the range of procedures amenable to a minimally invasive approach. Surgical difficulty assessment is essential for optimal patient selection, particularly during the surgeon's learning curve. Although several difficulty scoring systems (DSS) have been proposed, most have been developed for laparoscopic liver surgery. Therefore, their predictive performance in robotic series remains uncertain. Patients undergoing robotic liver resection for various indications were identified from a prospectively-maintained database. The Tampa, IWATE, Halls, Hasegawa, and Kawaguchi DSS were calculated based on variables available at the time of surgery. The predictive accuracy of the five DSS when assessing intraoperative and postoperative outcomes were compared using the ROC-AUC. Among 187 patients, unplanned conversion occurred in 4.3% (n = 8), severe morbidity in 14.4% (n = 27) and 90-day mortality in 3.2% (n = 6). Textbook Outcome in Liver Laparoscopic Surgery (TOLLS) was achieved in 66.8% (n = 125). The categorization of patients into surgical difficulty categories varied drastically among the DSSs. The Tampa DSS demonstrated the highest discriminatory ability for predicting unplanned conversion (AUC = 0.735), severe complication (AUC = 0.743), 90-day mortality (AUC = 0.722), and TOLLS achievement (AUC = 0.702) (Fig. 1). In this robotic series, the Tampa DSS showed superior performance in predicting postoperative outcomes. Standardized assessment of surgical complexity is crucial for accurate patient selection, reliable benchmarking, and tailored implementation of learning curves.