ABSTRACT
OBJECTIVE: To create a prognostic model determining the risk of tension pneumothorax and the need for intraoperative installation of additional drainage after thoracoscopic lobectomy. MATERIAL AND METHODS: A retrospective multiple-center study included patients who underwent thoracoscopic lobectomy for lung cancer between 2016 and 2022. One drainage tube was used after surgery in all cases. We synthesized data to expand patient selection using the Riley method and machine learning algorithm. In total, treatment outcomes in 1458 patients were analyzed. After identifying significant factors, we performed binary logistic regression analysis using backward stepwise inclusion of variables in accordance with the Akaike information criterion. After validating the model using the Bootstrap method (400 iterations) and original data set, we created a nomogram determining scoring characteristics, linear predictors and risk of postoperative tension pneumothorax. RESULTS: The incidence of tension pneumothorax was 4.53% (n=66). The most significant variables associated with pneumothorax and the need for additional pleural drainage were adhesions, intraoperative lung suturing, unclear interlobar groove, enlarged intrapulmonary lymph nodes and chronic obstructive pulmonary disease (p<0.001). The model's C-index was 0.957, mean absolute calibration error - 0.6%, calibration curve slope - 0.959. A score of 26 indicated a 95% risk of postoperative pneumothorax. CONCLUSION: We developed a prognostic model for tension pneumothorax after thoracoscopic lobectomy. Nomogram makes it possible to make a decision on intraoperative installation of additional pleural drainage tube and prevent complications associated with postoperative lung collapse.
