Development and Validation of a Model to Predict Acute Kidney Injury in Hospitalized Patients With Cirrhosis

Acute kidney injury (AKI) is a common complication in hospitalized patients with cirrhosis which contributes to morbidity and mortality. Improved prediction of AKI in this population is needed for prevention and early intervention. We developed a model to identify hospitalized patients at risk for AKI.

Admission data from a prospective cohort of hospitalized patients with cirrhosis without AKI on admission (n = 397) was used for derivation. AKI development in the first week of admission was captured. Independent predictors of AKI on multivariate logistic regression were used to develop the prediction model. External validation was performed on a separate multicenter cohort (n = 308).

In the derivation cohort, the mean age was 57 years, the Model for End-Stage Liver Disease score was 17, and 59 patients (15%) developed AKI after a median of 4 days. Admission creatinine (OR: 2.38 per 1 mg/dL increase [95% CI: 1.47-3.85]), international normalized ratio (OR: 1.92 per 1 unit increase [95% CI: 1.92-3.10]), and white blood cell count (OR: 1.09 per 1 × 10/L increase [95% CI: 1.04-1.15]) were independently associated with AKI. These variables were used to develop a prediction model (area underneath the receiver operator curve: 0.77 [95% CI: 0.70-0.83]). In the validation cohort (mean age of 53 years, Model for End-Stage Liver Disease score of 16, and AKI development of 13%), the area underneath the receiver operator curve for the model was 0.70 (95% CI: 0.61-0.78).

A model consisting of admission creatinine, international normalized ratio, and white blood cell count can identify patients with cirrhosis at risk for in-hospital AKI development. On further validation, our model can be used to apply novel interventions to reduce the incidence of AKI among patients with cirrhosis who are hospitalized.