The learning curve of video-assisted mediastinoscopic lymphadenectomy for staging of non-small-cell lung carcinoma.

OBJECTIVES: The objective of this study was to define the learning process of video-assisted mediastinoscopic lymphadenectomy (VAMLA) by the assessment of consecutive procedural metrics.

METHODS: We conducted a single-centre retrospective observational study of all consecutive VAMLAs performed between 2011 and 2018 for the staging of non-small-cell lung carcinoma. Learning curves were assessed using non-risk adjusted cumulative observed minus expected (CUSUM) failure charts of complications. Boundary lines were defined by the acceptable and unacceptable complication rates of 4.5% and 15.0%. The Kruskal-Wallis test with post hoc analysis was used to assess trends in operation time and blood loss.

RESULTS: Two-hundred-thirty-six unique VAMLAs by 4 surgeons performing their first procedures were evaluated. CUSUM charts of surgeons A and B showed a typical learning curve with an initial incline, followed by a turning point towards lower complications rates after 16-17 cases, whereas surgeons C and D showed an average performance. The median time between consecutive VAMLAs was shorter for surgeons A and B (13.0 vs 28.5-38.0 days for surgeons C and D). Overcoming the learning curve, complication rates of surgeons A and B decreased from 19% to 3% and from 18% to 5%, respectively. Operation time and blood loss showed a significant improvement after, respectively, 81-100 and 61-80 procedures compared to the first 20 procedures.

CONCLUSIONS: VAMLA is a safe procedure to adopt and perform with acceptable complication rates from the first operation onward, regardless of the caseload. To overcome its learning curve, 16-17 cases are required, preferably at least 1 per 2 weeks.