BACKGROUND: The objective of this study was to better characterize prolonged air leak (PAL), defined as an air leak longer than 7 days, and to develop and validate a predictive model of this complication after pulmonary resection. METHODS: All lung resections entered in Epithor, the French national thoracic database (French Society of Thoracic and Cardiovascular Surgery), were analyzed. Data collected between 2004 and 2008 (n=24,113) were used to build the model using backward stepwise variable selection, and the 2009 data (n=6,813) were used for external validation. The primary outcome was PAL. Results of the predictive model were used to propose a score: the index of PAL (IPAL). RESULTS: Prevalence of PAL after pulmonary resection was 6.9% (n=1,655) in the development data set. In the final model, 9 variables were selected: gender, body mass index, dyspnea score, presence of pleural adhesions, lobectomy or segmentectomy, bilobectomy, bulla resection, pulmonary volume reduction, and location on upper lobe. In the development data set, the C-index was 0.71 (95% confidence interval [CI], 0.70 to 0.72). At external validation, the C-index was 0.69 (95% CI, 0.66 to 0.72) and the calibration slope (ie, the agreement between observed outcomes and predictions) was 0.874 (<1). A score chart based on these analyses has been proposed. The formula to calculate the IPAL is the following: gender (F=0; M=4)-(body mass index-24)+2×dyspnea score+pleural adhesion (no=0; yes=4)+pulmonary resection (wedge=0; lobectomy or segmentectomy=7; bilobectomy=11; bulla resection=2; volume reduction=14)+location (lower or middle lobe=0; upper=4). CONCLUSIONS: Surgeons can easily use the well-validated model to determine intraoperative preventive measures of PAL.
BACKGROUND: The objective of this study was to better characterize prolonged air leak (PAL), defined as an air leak longer than 7 days, and to develop and validate a predictive model of this complication after pulmonary resection. METHODS: All lung resections entered in Epithor, the French national thoracic database (French Society of Thoracic and Cardiovascular Surgery), were analyzed. Data collected between 2004 and 2008 (n=24,113) were used to build the model using backward stepwise variable selection, and the 2009 data (n=6,813) were used for external validation. The primary outcome was PAL. Results of the predictive model were used to propose a score: the index of PAL (IPAL). RESULTS: Prevalence of PAL after pulmonary resection was 6.9% (n=1,655) in the development data set. In the final model, 9 variables were selected: gender, body mass index, dyspnea score, presence of pleural adhesions, lobectomy or segmentectomy, bilobectomy, bulla resection, pulmonary volume reduction, and location on upper lobe. In the development data set, the C-index was 0.71 (95% confidence interval [CI], 0.70 to 0.72). At external validation, the C-index was 0.69 (95% CI, 0.66 to 0.72) and the calibration slope (ie, the agreement between observed outcomes and predictions) was 0.874 (<1). A score chart based on these analyses has been proposed. The formula to calculate the IPAL is the following: gender (F=0; M=4)-(body mass index-24)+2×dyspnea score+pleural adhesion (no=0; yes=4)+pulmonary resection (wedge=0; lobectomy or segmentectomy=7; bilobectomy=11; bulla resection=2; volume reduction=14)+location (lower or middle lobe=0; upper=4). CONCLUSIONS: Surgeons can easily use the well-validated model to determine intraoperative preventive measures of PAL.