OBJECTIVES

Acute kidney injury (AKI) is commonly encountered in patients hospitalized for decompensated cirrhosis and is associated with prolonged hospital stays, increased treatment burden, and even mortality. The present study aimed to determine the prevalence of and develop a predictive nomogram for AKI in patients with decompensated cirrhosis.

METHODS

This cross-sectional, double-center study involved 544 patients hospitalized with decompensated cirrhosis. Acute kidney injury was diagnosed using American Gastroenterological Association's guidelines with one more criterion: an increase in serum creatinine ≥ 0.3 mg/dL within 48 h or an increase in serum creatinine ≥ 50% compared to baseline serum creatinine or when the urine output is reduced below 0.5 mL/kg/h for > 6 h. We used the Bayesian model averaging method find the optimal model for predicting AKI. A predictive nomogram was also developed to enable risk prediction.

RESULTS

The overall AKI prevalence was 26.7% (95% Confidence interval [CI] 25.7-27.7). The optimal model for predicting AKI included diuretic therapy (odds ratio [OR]: 5.55; 95%CI 3.31-9.33), infection (OR: 2.06; 95%CI 1.31-3.22), ascites (OR: 3.20; 95%CT: 1.67-6.13), Child-Pugh group C (OR: 2.91; 95%CI 1.84-4.62), serum potassium (OR per 1 mmol/L increase: 1.62; 95%CI 1.25-2.1) and serum chloride (OR per 1 mmol/L decrease: 1.03; 95%CI 1.01-1.06). The area under the receiver operating characteristic curve was 0.8, with a 95%CI ranging from 0.75 to 0.84.

CONCLUSIONS

Acute kidney injury was relatively common among patients hospitalized for decompensated cirrhosis. A novel nomogram-including diuretic therapy, infection, ascites, Child-Pugh group C, serum potassium and, serum chloride, was helpful for the selective screening of AKI in patients with decompensated cirrhosis.