Influencing factors and prediction models for AKI in ascites patients with hepatitis B cirrhosis

Objective  To explore the predictive factors of acute kidney injury （AKI） in patients with hepatitis B cirrhosis plus ascites and establish a new prediction model.

Methods  Fom January 1, 2018 to December 31, 2022, the relevant clinical data were retrospectively reviewed for 1350 patients with hepatitis B cirrhosis admitted into Department of Infection of First Affiliated Hospital of Xinjiang Medical University. According to the diagnostic criteria for AKI of International Ascites Club in 2015, they were divided into two groups of AKI and non-AKI. According to the types of variables, the data differences between two groups were compared by χ², T and Mann-Whitney U. The risk factors for AKI were screened by binary Logistic regression and ROC curve was plotted for evaluating the diagnostic value of each risk factor for AKI. An end-stage liver disease model was utilized for predicting AKI and short-term outcomes.

Results  Among 446 patients with hepatitis B cirrhosis and ascites, 111 cases developed AKI. The results of multivariate Logistic regression analysis revealed that prothrombin activity （OR = 1.031, 95%CI: 1.005-1.058）, blood urea nitrogen （BUN） （OR = 1.305, 95%CI:1.175-1.450）, MELD score （OR = 1.181, 95%CI: 1.100-1.267） and moderate ascites （OR = 7.218, 95%CI: 2.170-24.008） were independent risk factors for AKI （P＜0.05）. In ROC curve, AUC value of four-party combined predictor was the greatest （AUC = 0.895）.

Conclusion  Prothrombin activity, BUN, MELD score and ascitic severity are independent risk factors for AKI in patients with hepatitis B cirrhosis plus ascites. And combined prediction model established by these four factors has a high predictive value.