The everlasting issue of prolonged air leaks after lobectomy for non-small cell lung cancer: A data-driven prevention planning model in the era of minimally invasive approaches.

BACKGROUND AND

OBJECTIVES

Prolonged air leaks (PAL) are the most frequent complication after lobectomy for non-small cell lung cancer, even in case of minimally invasive approaches. We developed a novel score to identify high-risk patients for PAL during minimally invasive lobectomy.

METHODS

A dedicated database was created. We investigated preoperative candidate features and specific intraoperative variables. Univariate and subsequent logistic regression analysis with bootstrap resampling have been used. Model performance has been assessed by reckoning the area under the receiver operating characteristics curve and the Hosmer-Lemeshow goodness of fit.

RESULTS

PAL (>5 days) occurred in 72 (15.69%) patients. Five variables emerged from the model. Each one was assigned a score to provide a cumulative scoring system: forced expiratory volume in 1 second below 86% (P = 0.004, 1.5 points), body mass index <24 ( P = 0.002, 1 point), active smoking ( P = 0.001, 1.5 points), incomplete fissures ( P = 0.004, 1.5 points), and adhesions ( P = 0.0001, 1 point). The new score provided a stratification into four risk classes.

CONCLUSIONS

The risk score incorporates either general or more specific variables, providing a risk stratification that could be readily applied intra- and postoperatively. Henceforth, specific technical and management measures could be properly allocated to curb PAL.