Abstract
BACKGROUND: Operative mortality rates are of great interest to surgeons, patients, policy makers, and payers as a metric for quality assessment. Thirty-day mortality and discharge mortality have been presumed to capture procedure-related deaths. However, many patients die after the 30-day mark or are transferred to other facilities or to home and die there, leading to the underreporting of surgically related deaths. We hypothesized that a longer period of observation would address these concerns and provide a more accurate measure of operative mortality. METHODS: We retrospectively reviewed institutional databases of patients undergoing resection for lung cancer, esophageal cancer, and mesothelioma. Mortality rates at 30 and 90 days were calculated with 95% confidence intervals (CIs). RESULTS: From 1999 to 2012, 7,646 surgical resections were performed: 6,119 for lung cancer, 1,258 for esophageal cancer, and 269 for mesothelioma. Among the different cancers and across operations, the additional mortality from day 31 to 90 (1.4%; 95% CI, 1.2% to 1.8%; n=111) was similar to that by day 30 (1.2%; 95% CI, 1.0% to 1.5%; n=95), resulting in overall 90-day mortality (2.7%; 95% CI, 2.3% to 3.1%; n=206) that was more than double the 30-day mortality. CONCLUSIONS: Among patients who have undergone operations for thoracic malignancies, mortality attributable to the operation occurs beyond the first 30 postsurgical days as well as after hospital discharge. Because cancer operations constitute a large portion of general thoracic surgery, we recommend national databases consider the inclusion of 90-day mortality in their data collection.