Evaluation of the Learning Curve Threshold in Robot-Assisted Lung Cancer Surgery: A Nationwide Population-Based Study.

Background: Recent publications suggest that the threshold for validation of the learning curve is 25 procedures. The aim of this study was to evaluate this threshold using another rarely used method, based on a composite quality indicator.

Methods: We included all patients from the French medico-administrative database receiving robot-assisted surgery for lung cancer, with a focus on hospitals performing at least 25 procedures over the period 2019-2022. For postoperative complication analysis, we used the Clavien-Dindo classification. We used the sequential probability ratio test to estimate the number of procedures at which a hospital achieved its learning curve.

Results: In France, the number of robotic-assisted procedures has risen steadily in the past few years: 195 in 2019 and 1567 in 2022 (overall, 3706 Robot-Assisted surgeries). The total number of patients with Clavien-Dindo classification > II was 833 (24.7%). Among the 28 hospitals performing at least 25 procedures, eight achieved their learning curve with thresholds ranging from 94 to 174 procedures, and the median was 110. Severe complications such as acute respiratory distress syndrome, respiratory failure, heart failure, acute ischemia of the lower limbs, or pulmonary embolism were significantly more frequent in the group of hospitals that did not validate the learning curve threshold.

Conclusions: This study suggests that the threshold of 25 procedures may not be sufficient to validate the robot-assisted surgery learning curve in lung cancer surgery. To significantly reduce postoperative complications, a hospital would need to perform 94 to 174 procedures to guarantee patient safety.