Date of Award

January 2017

Document Type


Degree Name

Medical Doctor (MD)



First Advisor

Daneil J. Boffa


THE MANAGEMENT OF OPERABLE LUNG CANCER IN THE NATIONAL CANCER DATABASE Joshua E. Rosen, Michelle C. Salazar, Brian N. Arnold, Daniel C. Thomas, Justin D. Blasberg, Anthony W. Kim, Frank C. Detterbeck, and Daniel J. Boffa. Section of Thoracic Surgery, Department of Surgery, Yale University School of Medicine, New Haven, CT.

This thesis describes four studies that examine various aspects of the care of non-small cell lung cancer (NSCLC) patients in the United States. All studies were performed utilizing data from the National Cancer Database (NCDB), a clinical tumor registry that captures approximately 70% of all incident cancer cases in the United States.

The first study characterizes the natural history of operable NSCLC, with the hypothesis that it would be superior to non-operable NSCLC. Cohorts of operable and non-operable patients were selected from the NCDB and their overall survival was analyzed using the Kaplan-Meier (KM) technique. The 5-year overall survival for untreated operable NSCLC was 10.1%, 7.3% and 4.9% for stage I, II, and IIIA disease respectively. Untreated operable NSCLC was associated with a superior survival compared to untreated non-operable NSCLC at all stages of disease (P<0.001). The natural history of operable NSCLC is still poor, but varies with stage and is superior to that of non-operable NSCLC.

The second study compared the efficacy of surgical lobectomy and stereotactic body radiotherapy (SBRT) for the treatment of stage I NSCLC in healthy patients, with the hypothesis that lobectomy would be associated with superior survival. Cohorts of “healthy” (i.e. no comorbidities, not considered medically inoperable) patients were selected from the NCDB and propensity-matched cohorts of SBRT and lobectomy patients were compared using the KM technique. Lobectomy was associated with a superior 5-year survival compared to SBRT (59% vs 29%, P < 0.001). These findings suggest that lobectomy leads to superior long-term survival over SBRT in patients with clinical stage I NSCLC whose health does not prohibit the use of surgery.

The third study examined risk factors for perioperative mortality and extended length of stay (eLOS, > 14 days) in NSCLC patients undergoing surgery. The NCDB was queried for NSCLC patients undergoing surgery, and those that experienced an adverse outcome were compared to those who did not using multivariable logistic regression modeling. Overall 30-day mortality rate was 3.4% and varied by surgical procedure. The frequency of adverse events after lung cancer surgery in the NCDB was found to be in line with that of other large databases.

The final study examines the relationship between facility discharge practices and readmission rates following lobectomy for NSCLC with the hypothesis that facilities that have a systematic practice of rapid discharge will have higher readmission rates. Facilities discharge practices were characterized by their median LOS relative to the median LOS for all patients in the same diagnosis year. Risk-standardized readmission rates (RSRRs) were calculated for each level of facility discharge practices (from rapid to slow) and it was found that facilities with a practice of rapid discharge had a significantly lower mean RSRR than those with slower discharge practices. In conclusion, it is possible for facilities to develop a practice of early discharge after lobectomy without increasing their rate of readmission.


This thesis is restricted to Yale network users only. This thesis is permanently embargoed from public release.