A model to predict inhospital mortality in patients with cirrhosis, ascites and hyponatremia

Background and objective Hypervolemic hyponatremia is a late complication of portal hypertension. Hyponatremia is associated with a higher mortality in hospitalized patients. In this study, we evaluated the risk factors for inhospital mortality and developed a mortality prediction model in patients with cirrhosis and hyponatremia. Methods Using the national inpatient sample data for years 2016 and 2017, we identified cirrhotic patients hospitalized with ascites and hyponatremia (n = 9153). We identified independent risk factors of inhospital mortality and developed a prediction model in a training group and assessed its accuracy in a validation group. To enhance the clinical utility, we further stratified patients into low-, intermediate-, and high-risk mortality risk groups using cutoff points selected by decision tree analysis. Results The inhospital mortality in our cohort was 10.2% (n = 846). Multivariable analysis showed that age at least 65 years, variceal bleeding, sepsis, coagulopathy, and acute-on-chronic liver failure (ACLF defined as two or more organ failures) were independent risk factors for mortality. The prediction model using these five risk factors had an AUROC of 0.80 [95% confidence interval (CI), 0.78-0.82] for the training data and 0.83 (95% CI, 0.80-0.86) for the validation data. The mortality risks in the low-, intermediate-, and high-risk groups were 4% (95% CI, 3-4), 29% (95% CI, 28-33), and 43% (95% CI, 37-50), respectively. Conclusion We have developed a clinically meaningful inhospital prognostic model with excellent discrimination that will enable clinicians to risk stratify hospitalized patients with hyponatremia, ascites, and cirrhosis.