Model of lung cancer surgery risk derived from a Japanese nationwide web-based database of 78 594 patients during 2014–2015†

OBJECTIVES Using data obtained from a Japanese nationwide annual database with web-based data entry, we developed a risk model of mortality and morbidity after lung cancer surgery. METHODS The characteristics and operative and postoperative data from 80 095 patients who underwent lung cancer surgery were entered into the annual National Clinical Database of Japan data sets for 2014 and 2015. After excluding 1501 patients, the development data set for risk models included 38 277 patients entering in 2014 and the validation data set included 40 317 patients entering in 2015. Receiver-operating characteristic curves were generated for the outcomes of mortality and composite mortality/major morbidity. The concordance index was used to assess the discriminatory ability and validity of the model. RESULTS The 30-day mortality and overall mortality rates, including in-hospital deaths, were 0.4% and 0.8%, respectively, in 2014, and 0.4% and 0.8%, respectively, in 2015. The rate of major morbidity was 5.6% in 2014 and 5.6% in 2015. Several risk factors were significantly associated with mortality, namely, male sex, performance status, comorbidities of interstitial pneumonia and liver cirrhosis, haemodialysis and the surgical procedure pneumonectomy. The concordance index for mortality and composite mortality/major morbidity was 0.854 (P < 0.001) and 0.718 (P < 0.001), respectively, for the development data set and 0.849 (P < 0.001) and 0.723 (P < 0.001), respectively, for the validation data set. CONCLUSIONS This model was satisfactory for predicting surgical outcomes after pulmonary resection for lung cancer in Japan and will aid preoperative assessment and improve clinical outcomes for lung cancer surgery.