Learning Curve of Robotic Atrial Septal Defect Repair: A Multidimensional CUSUM Analysis of 284 Consecutive Cases.

Four distinct learning phases were identified in robotic ASD repair using multidimensional CUSUM analysis.

• Retrospective analysis of 284 consecutive adult patients undergoing robotic ASD repair.
• CUSUM was applied to operative time, cardiopulmonary bypass time, cross-clamp time, ICU stay, and hospital stay.
• Learning phases were defined by inflection points of the CUSUM curves.

Technical proficiency was achieved at Case 158, with significant reductions in operative, bypass, and cross-clamp times.

• Operative time fell from 186.8 ± 43.3 min to 133.7 ± 27.7 min.
• Cardiopulmonary bypass and cross-clamp times also decreased significantly (p < 0.001).
• Competency required a substantial case volume.

ICU stay and transfusion requirements progressively declined across learning phases.

• The decline was progressive across the phases identified by CUSUM.
• ICU stay was one of the CUSUM-analysed outcome parameters.

Mortality remained low across all learning phases.

• Low mortality was observed throughout the 284-case series, including the early learning phase.

Optimisation of postoperative care lagged behind technical mastery.

• Improvements in ICU and hospital stay metrics followed, rather than coincided with, gains in intraoperative technical performance (operative, bypass, cross-clamp times).

The authors conclude that multidimensional CUSUM provides an objective, performance-based assessment of robotic ASD learning and supports structured training and longitudinal monitoring.

• Competency acquisition required substantial case volume.
• The authors emphasise the need for structured training and ongoing performance monitoring in robotic cardiac surgery.