A comparative study of four thoracic mortality scores

Background The percentage of patients in resectable stages at initial diagnosis of non-small cell lung cancer (NSCLC) raises due to better screening programs. Therefore, risk prediction models are becoming more critical. Here, we validated and compared four established scoring models, the Thoracoscore, Epithor, Eurloung 2, and the simplified Eurolung 2 (2b), in their ability to predict 30-day mortality. Methods All consecutive patients undergoing anatomical pulmonary resection were included. The performance of the four scoring systems was assessed with Hosmer-Lemeshow goodness-of-fit test (calibration) and receiver operating characteristic (ROC) curves (discrimination). We compared the area under the curve (AUC) of the ROC curves by DeLong's method. Results A total of 624 patients underwent surgery for NSCLC at our institution between 2012 and 2018 30-day mortality was 2.2% (14 patients). The AUC for Eurolung 2 and the simplified Eurolung 2 (0.82) were greater than those of the other scoring systems, Epithor (0.71) and Thoracoscore (0.65). In addition, the DeLong analysis showed a significant superiority of Eurolung 2 and Eurolung 2b over the Thoracoscore (p = 0.04); there were no significant differences compared to Epithor. Conclusion Eurolung 2 and the simplified Eurolung 2 were the favorable scoring systems for predicting 30-day mortality compared to Thoracoscore and Epithor. Therefore, we recommend using Eurolung 2 or the simplified Eurolung 2 for preoperative risk stratification.