Table 4.

Accuracy of different scoring systems in derivation and validation cohorts

Models	Cut-off	Sensitivity (%)	Specificity (%)	Youden	AUROC (95%CI)	P value	P value
							compared with GIMNS
Derivation cohort (n = 248)
GIMNS	0.4	73.5	78.7	0.522	0.830 (0.778-0.882)	< 0.001	-
MELD	24.8	61.4	76.6	0.380	0.759 (0.670-0.821)	< 0.001	0.029
CLIF-SOFA	9.0	72.7	68.1	0.408	0.767 (0.706-0.827)	< 0.001	0.045
COSSH-ACLF	7.5	56.8	84.0	0.408	0.759 (0.696-0.821)	< 0.001	0.026
Validation cohort (n = 123)
GIMNS	0.7	57.8	79.6	0.374	0.732 (0.642-0.821)	< 0.001	-
MELD	26.6	65.6	63.0	0.286	0.623 (0.520-0.726)	< 0.050	0.013
CLIF-SOFA	11.0	29.7	94.4	0.241	0.661 (0.563-0.758)	< 0.010	0.049
COSSH-ACLF	7.9	51.6	81.5	0.331	0.674 (0.577-0.770)	< 0.001	0.122

AUROC: Area under the receiver operating characteristic curve; CLIF-SOFA: Chronic liver failure-sequential organ failure assessment; COSSH-ACLF: Chinese Group on the study of severe hepatitis B-acute-on-chronic liver failure; MELD: Model for end-stage liver disease; CI: Confidence interval.