Predicting risk of intensive care unit admission after resection for non-small cell lung cancer: a validation study

A model for predicting the risk of emergency, unplanned intensive care unit (ICU) admission after lung resection for lung cancer has been reported. However, it has not been validated outside of the derivation cohort, and the aim of our study was to undertake external validation at our institution. We reviewed a series of consecutive patients who underwent major lung resection for non-small cell lung cancer over a 6-year period. Test performance was evaluated by area under the receiver operator characteristic (ROC) curve. Between 2003 and 2008, 425 patients underwent major lung resections for lung cancer. The mean age (SD) was 65 (10) years and 241 (57%) were men. A total of 77 (18%) patients were admitted to ICU, 47 for elective admission and 30 (7%) for treatment of post-procedure complications. Of the 30 patients admitted for complications, the median length of ICU (interquartile range) stay was 3 days (1-15 days). The mortality rate among these patients was 17%. The area under the ROC curve was 0.66 (95% CI 0.53-0.79). The Brunelli scoring system had moderate discriminating ability to predict the risk of ICU admission after lung resection in our institution.