Georg Wiltberger1, Babett Muhl2, Christian Benzing3, Georgi Atanasov3, Hans-Michael Hau2, Matthias Horn4, Felix Krenzien3, Michael Bartels2. 1. Department of Visceral-, Transplantation-, Thoracic- and Vascular Surgery, University Hospital Leipzig, Leipzig, Germany. Electronic address: Georg.Wiltberger@medizin.uni-leipzig.de. 2. Department of Visceral-, Transplantation-, Thoracic- and Vascular Surgery, University Hospital Leipzig, Leipzig, Germany. 3. Department of General-, Visceral- and Transplant Surgery, Charité University Medical Department Berlin, Campus Virchow, Berlin, Germany. 4. Institute for Medical Informatics, Statistics and Epidemiology (IMISE), University of Leipzig, Leipzig, Germany.
Abstract
INTRODUCTION: Morbidity after pancreaticoduodenectomy (PD) remains a major concern with high rates. The aim of this study was to identify preoperative risk factors and create a new risk score to predict major complications after PD. METHODS: Medical records of patients undergoing PD between 1993 and 2014 were retrospectively reviewed according to survival and surgical and non-surgical complications. A split-sample cross validation was conducted in which the original cohort was randomly selected to a modelling and a validation group at a ratio of 2:1. Univariate and multivariate analysis were carried out on the modelling set to identify preoperative risk factors, which were entered into a binary logistic regression model with stepwise backward elimination to develop the risk score model. Receiver operating curve analysis was implemented to judge the model's prediction ability. RESULTS: PD was performed in 405 patients. A total of 29.1% (118 patients) developed major complications. On multivariate analysis, American Society of Anaesthesiologists (ASA) score and obesity as well as the presence of cardiovascular and pulmonary comorbidities were significant predictors for major complications. A risk score was derived from the regression model and successfully tested on the validation set (area under the curve = 0.84). CONCLUSION: The risk score showed a high accuracy to predict major complications after PD based on preoperative parameters only. This simple and quick approach allows for individualized risk assessment and may improve preoperative counselling and patient selection for perioperative treatment strategies.
INTRODUCTION: Morbidity after pancreaticoduodenectomy (PD) remains a major concern with high rates. The aim of this study was to identify preoperative risk factors and create a new risk score to predict major complications after PD. METHODS: Medical records of patients undergoing PD between 1993 and 2014 were retrospectively reviewed according to survival and surgical and non-surgical complications. A split-sample cross validation was conducted in which the original cohort was randomly selected to a modelling and a validation group at a ratio of 2:1. Univariate and multivariate analysis were carried out on the modelling set to identify preoperative risk factors, which were entered into a binary logistic regression model with stepwise backward elimination to develop the risk score model. Receiver operating curve analysis was implemented to judge the model's prediction ability. RESULTS: PD was performed in 405 patients. A total of 29.1% (118 patients) developed major complications. On multivariate analysis, American Society of Anaesthesiologists (ASA) score and obesity as well as the presence of cardiovascular and pulmonary comorbidities were significant predictors for major complications. A risk score was derived from the regression model and successfully tested on the validation set (area under the curve = 0.84). CONCLUSION: The risk score showed a high accuracy to predict major complications after PD based on preoperative parameters only. This simple and quick approach allows for individualized risk assessment and may improve preoperative counselling and patient selection for perioperative treatment strategies.
Authors: Nina L Eng; Danielle E Mustin; Brendan P Lovasik; Michael K Turgeon; Adriana C Gamboa; Mihir M Shah; Kenneth Cardona; Juan M Sarmiento; Maria C Russell; Shishir K Maithel; Jeffrey M Switchenko; David A Kooby Journal: Ann Surg Oncol Date: 2020-07-20 Impact factor: 5.344
Authors: Stine Dam Henriksen; Poul Henning Madsen; Anders Christian Larsen; Martin Berg Johansen; Inge Søkilde Pedersen; Henrik Krarup; Ole Thorlacius-Ussing Journal: Oncotarget Date: 2017-09-30