Personalized prediction of acute kidney injury in patients with pancreatic necrosis

Abstract

Relevance. The incidence of acute pancreatitis is growing worldwide, being one of the leading causes of hospitalization in urgent surgery. The most common complication of pancreatic necrosis (PN) in the aseptic phase is acute kidney injury (AKI), which is an independent risk factor for an unfavorable outcome.The objective was to develop a personalized risk model for AKI in the aseptic phase of pancreatic necrosis.Materials and methods. A comparative cohort study of the results of treatment of 502 patients with pancreatic necrosis was conducted. The primary endpoint was considered to be the development of AKI, for the development of a personalized model of the probability of its development in sterile pancreatic necrosis, binary logistic regression analysis was used.Results. A model of independent variables was developed that reliably (p < 0.001) determined that with an increase in age by 1 year, the probability of developing AKI increased by 2.3%, and with a history of chronic kidney disease in a patient – by 3.2 times.The same model demonstrates that the risk of AKI in patients with pancreatic necrosis with an increase in glomerular filtration rate by 1 ml·min–1·1.73 m2 and with the use of balanced crystalloid solutions decreased by 5.0% and 3.0 times, respectively.The specificity of the model was 79.8%, sensitivity – 79.1%.Conclusion. The proposed model makes it possible to reliably predict the individual risk of AKI on the first day of hospitalization.