|
Back to 2015 Annual Meeting Program
Pathologic Diagnosis Is an Independent Predictor of Complications in Pancreatectomy
Tyler J. Mouw, Sarah L. Mott, James J. Mezhir*
Surgery, Univ of Iowa, Iowa City, IA
Background: Previous reports from single institutions demonstrate that pancreatic duct diameter and texture are risk factors for complications in pancreatectomy. As surgeons we infer that patients with certain pathologic subtypes that do not result in significant changes in the pancreas would result in a higher rate of complications. Data are needed to determine which specific pathologic subtypes would pose the highest risk for complications for optimal perioperative management, especially in the development of protocol-based systems for all pancreatectomy patients.
Methods: The ACS-NSQIP dataset was evaluated for patients treated with pancreatectomy from 2005-2012. All perioperative variables were included for evaluation of complications and mortality. Post-operative pathologic diagnosis was grouped based on clinical significance. Primary endpoints were any complication, wound/infectious complication, bleeding/clotting, and mortality. To strengthen the predictability of the multivariate model, the dataset was randomly split to create model development and validation cohorts.
Results: 15,218 patients were included in the analysis; 10,036 patients were treated with pancreaticoduodenectomy, 4,826 with distal pancreatectomy, and 356 total pancreatectomy. There were 5,505 patients (36.2%) who experienced a complication and 358 patients died within 30 days (2.4% mortality rate). The rate of wound or infectious complications was 29.0% and bleeding/clotting complication, 16.5%. The c-statistic for model performance was above 0.5 for all outcome measures. There were multiple well-known predictors of complications and mortality such as poor performance status, transfusion, BMI, OR time, and abnormal lab values.
Pathologic diagnosis was evaluated and revealed differences with respect to complications on multivariate analysis. A pathologic diagnosis of bile duct/ampullary tumor (OR 1.23, p<0.01), or duodenal tumors (OR 1.43, p<0.01) was in independent predictor of overall complications compared with ductal adenocarcinoma, benign cystic neoplasms, pancreatitis, and neuroendocrine tumors. These patients also had higher rates of wound/infectious complications: bile duct/ampullary (OR 1.57, p<0.01) and duodenal tumors (OR 1.69, p<0.01). A diagnosis of benign cystic neoplasm was predictive of a lower risk for bleeding/clotting complications (OR 0.74, p<0.01). Pathologic diagnosis did not predict mortality in this study.
Conclusions: In both primary and validation cohorts from ACS-NSQIP, pathologic diagnosis of bile duct/ampullary and duodenal tumors independently predicted increased overall complications and wound/infectious complications following pancreatectomy; benign cystic neoplasms were at lower risk for bleeding/clotting complications. These data may be used to tailor pancreatectomy protocols in select patients being treated with pancreatectomy.
Back to 2015 Annual Meeting Program
|
