How Does a Robotic Lung Resection Programme Evolve? Learning Curve and Conversion Analysis of the Versius CMR...

Background/Objectives: Robotic-assisted thoracic surgery continues to spread internationally for anatomical lung resection, yet evidence on learning curve dynamics and conversion behaviour remains limited beyond the da Vinci system. The published clinical experience on the Versius CMR Surgical System does not yet match that of the da Vinci system. This study examines the learning curve and conversion pattern of a mature Versius programme using cumulative sum (CUSUM) analysis. Methods: Single-centre retrospective cohort study encompassing 206 consecutive anatomical lung resections (154 lobectomies, 52 segmentectomies) undertaken using the Versius CMR platform between April 2023 and February 2026. CUSUM analysis was applied at programme and individual surgeon level. Logistic regression was performed to identify independent predictors of conversion. Operative time and post-operative length of stay were examined, along with a comparative sub-analysis between converted and pure robotic cases, stratified by surgeon. Results: The programme-level CUSUM inflection point occurred at case 49 (7.3 months), with mean operative time falling from 250.9 ± 65.4 min in the early phase to 215.5 ± 46.8 min in the late phase (p = 0.001). Post-operative LOS improved from 5.8 to 4.7 days (p = 0.031). Forty-six cases (22.3%) required conversion, 87% of which were elective, with no significant change across programme phases (p = 1.00). A six-fold difference in conversion rate existed between the two main surgeons (35.4% vs. 6.1%). Surgeon-1's conversions carried no operative time penalty (225.2 vs. 223.3 min, p = 0.96) or LOS penalty (3.9 vs. 5.0 days, p = 0.83), whereas Surgeon-2's converted cases were significantly longer (333.3 vs. 216.9 min, p = 0.004) and associated with longer hospital stay (8.8 vs. 4.7 days, p = 0.004). On multivariable analysis, surgeon identity (aOR 5.39, 95% CI 2.25-12.92, p < 0.001), intra-operative adhesions (aOR 3.41, 95% CI 1.53-7.64, p = 0.003) and clinical stage II or higher (aOR 3.55, 95% CI 1.43-8.79, p = 0.006) were each independently associated with conversion. Conclusions: The Versius CMR system supports a definable and reproducible learning curve in anatomical lung resection. Conversion was determined both by operative difficulty and by individual surgeon decision-making philosophy, the latter accounting for the greater share of the programme's conversion burden. Early elective conversion, when indicated, preserves operative efficiency and post-operative recovery, and should be viewed as wise surgical judgement, not a marker of programme immaturity.