Table 2.

Features of illustrative systems for mortality prediction in acute liver failure

Study	Publication year	Criteria	Era	Etiology	n	Non-survivorsa	Sensitivity	Specificity	AUROC	Comment
McPhail et al. [20]	2016	KCC	2001–2015	All	2153	Not Given	0.59 (0.56–0.62)	0.79 (0.77–0.81)	0.76	Meta-analysis
Cholongitas et al. [25]	2012	SOFA	1993–2010	Paracetamol	125	58 (46%)	0.67	0.80	0.79	SOFA Score threshold 12
McPhail et al. [20]	2016	MELD	2001–2015	All	2153	Not given	0.74 (0.71–0.77)	0.67 (0.64–0.69)	0.78	Meta-analysis
Bechmann et al. [29]	2010	MELD/CK18	2006–2009	All	68	18 (27%)	0.85 (0.69–1.0)	0.76 (0.6–0.91)	0.94	Peak M65 fragment
Koch et al. [22]	2016	ALFSG	1998–2013	All	1974	987 (50%)	Not given	Not given	0.84	Prediction of survival not death
Rutherford et al. [30]	2012	ALFSG/CK18	1998–2011	All	500	251 (50%)	0.86	0.65	0.82	Admission M30 fragment
Antoine et al. [32]	2012	Acetylated HMGB1	Not Given	Paracetamol	78	27 (35%)	Not given	Not given	0.87	Admission values

KCC Kings College Criteria, SOFA sequential organ failure assessment, MELD model for end-stage liver disease, CK18 cytokeratin 18, ALFGSG Acute liver failure study group, HMGB1 high-mobility group box-1

^aNon-survivors; cases who died or were transplanted. Figures in parentheses are 95% Confidence Intervals where given