readmissions after pulmonary resections. We examined the NSQIP database to provide baseline data for quality improvement efforts.

METHODS: The ACS-NSQIP dataset for years 2012-2013 was investigated. Patients who underwent lung resections were identified by CPT code, stratified into video-assisted thoracic surgery (VATS) and THOR groups, and further stratified by ICD9 into cancer vs non-cancer. Outcomes included 30-day morbidity and mortality in addition to rates and reason for readmission. RESULTS: We identified 8,930 lung resection patients; 59.5% of resections were performed for lung cancer and of these, 73% were lobectomy. Overall, 56% of cases were performed using VATS. Morbidity and mortality rates were 9% and 1.2% for VATS vs 15.3% and 2% for THOR (p<0.001). Median length of stay was 4 days for VATS (interquartile range [IQR] 2-6 days), and 5 days (IQR 4-8 days) for THOR (p<0.001). Unplanned readmission occurred within 30 days in 7.4% of patients and was not dependent on surgical approach or preoperative diagnosis. The reason for readmission was more likely related to air leak in the VATS group vs infections in the THOR group. Multivariate predictors of readmission included American Society of Anesthesiologists (ASA) 3, pneumonectomy, and the occurrence of postoperative complications as shown in the table.