Establishment of prognostic scoring models for different etiologies of acute decompensation in hospitalized patients with cirrhosis.

Objective: Acute decompensation (AD) in liver cirrhosis has high mortality. We assessed prognostic scoring models and established prediction models for different etiologies of AD.

Methods: This retrospective analysis included 732 patients hospitalized with acute decompensated cirrhosis without acute-on-chronic liver failure. We performed logistic regression analysis of risk factors for mortality associated with different etiologies, to establish predictive models.

Results: Patients with different etiologies, scored using different scoring systems and various impact factors, exhibited differences with respect to mortality. MELD, CLIF-C-AD, MELD-Na, and AARC-ACLF scores exhibited adequate predictive ability for mortality. Area under the receiver operating characteristic curve for 28-day mortality for MELD, CLIF-C-AD, MELD-Na, AARC-ACLF, and the newly developed AD scores was 0.663, 0.673, 0.657, 0.662, and 0.773, respectively, in the hepatitis B virus group (HBV-AD score =−5.51 + 0.07*WBC count (10/L) +0.7*AD sum+0.4*AARC-ACLF score); 0.731, 0.737, 0.735, 0.689, and 0.778, respectively, in the alcoholic liver disease group (ALD-AD score =−4.55 +0.08* WBC count (10/L) +1.34* AD sum); and 0.765, 0.767, 0.814, 0.720, and 0.814, respectively, in the Others group (OTHERS-AD score =−2.14 + 1.24*MELD-Na score +4.49*AD sum).

Conclusions: The newly developed scoring models for short-term mortality were superior to the other scoring systems in predicting prognosis of acute decompensated cirrhosis in hospitalized patients.