Differential effects of operative complications on survival after surgery for primary lung cancer.

Objective: Complications adversely affect survival after lung cancer surgery. We tested the hypothesis that effects of complications after lung cancer surgery on survival vary substantially across the spectrum of postoperative complications.

Methods: The Society of Thoracic Surgeons General Thoracic Surgery Database was linked to Medicare data for lung cancer resections from 2002 through 2013. Linkage was achieved for 29,899 patients. A survival model was created that included operative complications as explanatory variables and adjusted for relevant baseline covariates. Because of violation of the proportional hazard assumption, we used time-varying coefficient Cox modeling for the complication variables.

Results: Median patient age was 73 years, and 48% were male. Procedures performed were lobectomy in 69%, wedge in 17%, segmentectomy in 7%, bilobectomy in 3%, pneumonectomy in 3%, and sleeve lobectomy in 1%. Most frequent complications were atrial arrhythmia (14%), pneumonia (4.3%), reintubation (3.8%), delirium (2%), and acute kidney injury (1.4%). In the early period (0-90 days), 12 complications are associated with worse survival. From 3 to 18 months after surgery, only 4 complications are associated with survival: delirium, blood transfusion, reintubation, and pneumonia. After 18 months, only sepsis and blood transfusion are associated with a significant late hazard.

Conclusions: Our analysis confirmed the presence of differential magnitude and time-varying effects on survival of individual complications after lung cancer surgery. We conclude that the derived time-dependent hazard ratios can serve as objective weights in future models that enhance performance measurement and focus attention on prevention and management of complications with greatest effects.