Table 3.

Description of models to predict risk of mortality at several time points and their predictive ability in the CANONIC (discovery set) and the PREDICT (validation set) study cohorts

Model	C-index
	At 7 days	At 14 days	At 28 days	At 60 days	At 90 days
Discovery set: CANONIC study cohort (n=831)
% of deaths	2.41%	5.78%	9.51%	15.52%	18.53%
CLIF-C MET 1	0.927	0.900	0.842	0.805	0.784
CLIF-C MET 2	0.937	0.921	0.853	0.808	0.784
MELDNa score	0.810*†	0.813*†	0.788*†	0.779	0.772
Validation set: PREDICT study cohort (n=851)
% of deaths	1.18%	4.58%	8.81%	15.28%	18.21%
CLIF-C MET 1	0.940	0.834	0.791	0.780	0.771
CLIF-C MET 2	0.947	0.857	0.810	0.788	0.777
MELDNa score	0.793*†	0.773†	0.755†	0.713*†	0.700*†

Set of variables included initially in model 1: log₂(4-hydroxy-3-methoxyphenylglycol sulfate, log₂(hexanoylcarnitine), log₂(L-Saccharopine), log₂(N-Acetyl-aspartil-glutamate), log₂(p-Hydroxyphenyllactic acid), log₂(D-Galacturonic acid), log₂(N-Acetyl-L-alanine), log₂(butyrylcarnitine), log₂(cystathionine), log₂(octanoylcarnitine) and age. Model 2 also included creatinine, bilirubin, INR and white cell count.

*P value <0.01 compared with the metabolomic model 1 for differences larger than 0.05.

†P value <0.01 compared with the metabolomic model 2 for differences larger than 0.05.

CLIF-C ACLF, Chronic Liver Failure Consortium ACLF; INR, international normalised ratio.