Learning curves in Versius-assisted colorectal surgery: a risk-adjusted CUSUM analysis from a single center

To evaluate procedure- and operator-specific learning curves in Versius-assisted colorectal cancer surgery using risk-adjusted cumulative sum (RA-CUSUM) analysis in a real-world single-center setting.

This retrospective study included consecutive adult patients with histologically confirmed colorectal cancer who underwent elective robotic surgery using the Versius system between December 2022 and August 2025. Right hemicolectomy (RHC), sigmoid resection (SR), and anterior rectal resection (AR) were analyzed. Learning curves were assessed using RA-CUSUM analysis of skin-to-skin operative time, adjusted for patient- and procedure-specific variables. Separate models were constructed for each procedure and operator.

A total of 156 procedures were included: 79 AR, 51 RHC, and 26 SR. Median operative time was 215 min (IQR 185-255). Conversion to open surgery was low (3.2%), and major postoperative complications were infrequent. RA-CUSUM analysis demonstrated distinct procedure- and operator-specific learning patterns. RHC and SR reached earlier stabilization, whereas AR required a higher case volume to achieve consistent operative time reduction, with more prolonged variability in the learning phase. Oncological quality indicators were within accepted standards.

Learning curves in Versius-assisted colorectal surgery are procedure- and operator-dependent. RA-CUSUM provides a structured method for monitoring trends in operative performance and may support structured training pathways and quality monitoring during implementation of robotic surgery programs.