Abstract
BACKGROUND: Early chest tube removal should be considered to enhance recovery after surgery. The current study aimed to provide a predictive algorithm for air leak episodes (ALE) and to create a knowledge base for early chest tube removal. METHODS: This retrospective study enrolled patients who underwent thoracoscopic anatomical pulmonary resections in our unit. We defined ALE as any airflow ≥ 10 mL/min recorded in the follow-up charts based on the digital thoracic drainage device. Multivariate regression analysis was used to control for preoperative and intraoperative confounding factors. The ALE prediction algorithm was constructed by combining an additive ALE risk-scoring system using the coefficients of the significant predictive factors with the intraoperative water-sealing test. RESULTS: In 485 consecutive thoracoscopic major pulmonary resections, ALE developed in 209 (43%) patients. Statistically significant ALE-associated preoperative factors included male sex, lower body mass index, radiologically evident emphysema, lobectomy, and upper lobe surgery. Significant ALE-associated intraoperative factors were incomplete fissure and pleural adhesion. The ALE risk scoring demonstrated an average area under the receiver operating characteristic curve of 0.72 in the fivefold cross-validation test. The ALE prediction algorithm correctly predicted ALE-absent patients at a negative predictive value of 80%. CONCLUSIONS: The algorithm may promote the optimization of the chest tube-dwelling duration by identifying potential ALE-absent patients for accelerated tube removal.