Background: The optimal strategy for imaging surveillance of non-small cell lung cancer (NSCLC) patients after curative intent surgery is unknown. Current guidelines recommend computed tomography (CT) every 6-12 months for 2 years and then annually. There are no large population-based studies identifying how patients are managed, or comparing the effectiveness of chest radiography (CXR) and CT surveillance. Methods: We performed a retrospective cohort study using Surveillance, Epidemiology and End Results (SEER)-Medicare data to determine the primary surveillance modality following surgical resection in stage I-IIIA NSCLC between 1998 and 2009. Primary surveillance modality was defined as the imaging study used between 90-365 days after surgery. Comparative effectiveness of CT vs. CXR surveillance was explored in terms of overall survival (OS) using a stratified Cox model based on stage and adjusted for age, gender, race, CMI, Charlson comorbidity index, and adjuvant chemotherapy. Results: 5,968 (54%)patients were followed by CT, and 5,083 (46%) by CXR. Patients with earlier stage, older age, and lower census median income (CMI) were less likely to undergo CT surveillance (p<0.001). CT surveillance increased over the study period from 23% in 1998 to 68% in 2009 (p<0.001). In the analysis of surveillance modality and OS, a significant interaction was identified between imaging and diagnosis year (p<0.001). The effect of CT surveillance on OS steadily improved over time, and was significantly better than CXR in the most recent time periods of study (Table). Conclusions: OS was improved in patients with CT surveillance in the most recent time periods of analysis supporting surveillance guidelines. Further studies to determine how CT surveillance leads to improved outcomes, to evaluate the appropriate interval of CT imaging, and to elucidate why patients are not followed according to guidelines are warranted.

*p<0.001