Validation of Risk Score in Predicting Early Readmissions in Decompensated Cirrhotic Patients: A Model Based on the Administrative Database

Abstract

Early readmission in patients with decompensated liver cirrhosis leads to an enormous burden on health care use. A retrospective cohort study using the 2013 and 2014 Nationwide Readmission Database (NRD) was conducted. Patients with a diagnoses of cirrhosis and at least one feature of decompensation were included. The primary outcome was to develop a validated risk model for early readmission. Secondary outcomes were to study the 30-day all-cause readmission rate and the most common reasons for readmission. A multivariable logistic regression model was fit to identify predictors of readmissions. Finally, a risk model, the Mumtaz readmission risk score, was developed for prediction of 30-day readmission based on the 2013 NRD and validated on the 2014 NRD. A total of 123,011 patients were included. The 30-day readmission rate was 27%, with 79.6% of patients readmitted with liver-related diagnoses. Age <65 years; Medicare or Medicaid insurance; nonalcoholic etiology of cirrhosis; ≥3 Elixhauser score; presence of hepatic encephalopathy, ascites, variceal bleeding, hepatocellular carcinoma, paracentesis, or hemodialysis; and discharge against medical advice were independent predictors of 30-day readmission. This validated model enabled patients with decompensated cirrhosis to be stratified into groups with low (<20%), medium, (20%-30%), and high (>30%) risk of 30-day readmissions. Conclusion: One third of patients with decompensated cirrhosis are readmitted within 30 days of discharge. The use of a simple risk scoring model with high generalizability, based on demographics, clinical features, and interventions, can bring refinement to the prediction of 30-day readmission in high-risk patients; the Mumtaz readmission risk score highlights the need for targeted interventions in order to decrease rates of readmission within this population.