Table 2.

Sensitivity and specificity of scoring systems for alcoholic hepatitis: sensitivities (Sen)/specificities (Spec)

Study	Patients	Study	Predictive	MELD	DF	CTP	Glasgow	Conclusions
		design	Mortality	Sen/Spec (%)	Sen/Spec (%)	Sen/Spec (%)	Sen/Spec (%)
Sheth et al[21] 2002	34	Retrospective	30 d	≥ 11	≥ 32	N/A	N/A	MELD equivalent to DF
				86/82	86/48
Kulkarni et al[17] 2004	41	Retrospective	28 d	N/A	≥ 33	N/A	N/A	DF ≥ 32 is appropriate.
					66.7/61.5			High mortality in DF < 32
Dunn et al[22] 2005	73	Retrospective	90 d	≥ 21	≥ 37	N/A	N/A	MELD equivalent to DF
				75/75	88/65
Srikureja et al[23] 2005	202	Retrospective	Not given	Admission: ≥ 18	≥ 32	≥ 12	N/A	Admission MELD equivalent
				85/84	83/84	76/80		to DF
				Wk 1: ≥ 20
				91/85
Forrest et al[16] 2005	134	Retrospective	28 d	N/A	≥ 32	N/A	≥ 9	GAHS more accurate in
			84 d		28 d		28 d	predicting mortality compared
					96/27		81/61	to DF
					84 d		84 d
					95/31		78/66
Forrest et al[16] 2005	46	Retrospective	28 d	≥ 11	N/A	N/A	≥ 9	GAHS more accurate in
			84 d	28 d			28 d	predicting 84 d mortality
				92/29			75/68	GAHS equivalent to MELD in
				84 d			84 d	predicting 28 d mortality
				92/29			69/67