Beware of the plateau trap: a multidimensional analysis redefining the learning curve of robotic lobectomy in non-small cell lung cancer.

Current definitions of the learning curve for robotic lobectomy are derived from small cohorts and lack standardized criteria for proficiency. Whether these series can accurately distinguish true surgical mastery from improved operative efficiency remains uncertain.We analyzed 300 consecutive patients with non-small cell lung cancer who underwent robotic portal lobectomy with four arms (RPL-4) between June 2018 and July 2023, using a prospectively maintained database. Learning phases were determined using cumulative sum and multivariable analyses. Perioperative outcomes were compared across phases. Three phases were identified: learning (cases 1-81), plateau (82-137), and mastery (> 137). Median console time declined significantly across phases (88.0 vs. 69.0 vs. 66.0 min, P < 0.001). Median blood loss also decreased (50.0 vs. 50.0 vs. 40.0 mL, P < 0.001). Despite improved efficiency, postoperative complication rates remained elevated during the plateau phase (10.7%), primarily due to Clavien-Dindo Grade II events (8.9%), before decreasing to 3.1% in the mastery phase (P = 0.016). Furthermore, sustained reductions in chest tube duration and length of stay were observed only after > 137 cases. RPL-4 demonstrates a prolonged 3-phase learning curve with a deceptive plateau. During this phase, gains in efficiency obscure persistently high postoperative complication rates, leading to underestimation of the true learning process. Training benchmarks should prioritize comprehensive surgical quality rather than operative speed alone.