PROSPECTS OF EARLY COMPLICATION AFTER PANCREATICODUODENECTOMY PREDICTION

Abstract

Pancreaticoduodenectomy (PD) is an operation with a significant number of complications, ranging sometimes from 30 to 60%. The aim of the study was to develop a mathematical model for preoperative determination of the risk of postoperative complications in patientsthat are considered as candidates for PD. The retrospective study of 108 case histories of patients who underwent PD in the period from 2008 to 2021 at the Surgery Clinic No. 2 at the Dnipro Regional Hospital named after I.I. Mechnikov was executed. Various postoperative complications occurred in 43 (39.8%) patients. There were on average 2 different complications in one patient. Factors that significantly influenced the development of postoperative complications were identified, namely: the presence of diseases of the cardiovascular system (rs = 0,2; p≤0,05), the volume and severity of blood loss during surgery (rs = 0,27; p≤0,05 and rs = 0,29; p≤0,05); the density of the pancreas parenchyma in the region of anastomosis formation (determined by palpation) (rs = 0,28; p≤0,05); diameter of the Wirsung duct (rs = - 0,21; p≤0,05). To build a logit model, it was decided to use the factors that can be determined before the operation: the diameter of the Wirsung duct; the density of the pancreas parenchyma; the presence of comorbidities (cardiovascular diseases). Since the presence of pancreatic fistula significantly affected the total number of complications (rs = 0,53; p≤0,05), it was decided to add to the model BMI, which was a significant risk factor for these complications (rs = 0,2; p≤0.05). ROC-analysis was performed to determine the cut-off points of quantitative indicators and their ability to predict the occurrence of complications in the postoperative period. The logit-regression equation obtained during the calculations was evaluated by the criterion χ2 , which was 13,48 (p=0,019), as well as by the ROC- analysis. The area under the ROC curve was AUC=0,7, 95% SI (0.602-0.786), sensitivity – 81,7%, specificity – 50,0%, p<0,001), which indicates good operational characteristics of the model. The cut-off point was ≥ 0,5, i.e. the obtained model allows to distinguish 2 variants of risk of postoperative complications - high and low.

Keywords

• pancreatoduodenal resection
• risk factors
• logit-regression.