Liver transplantation in the critically ill: a multicenter Canadian retrospective cohort study

Table 3 Multivariable analysis: predictors of 90-day mortality in 198 critically ill cirrhosis patients who underwent liver transplantation

Covariate	Unadjusted	Model 1 (n= 180)	Model 2 (n= 110)	Model 3 (n= 140)
Age	1.06 (1.01-1.11)^a	1.07 (1.01-1.14)^b	1.12 (1.02-1.22)^c	1.07 (1.01-1.14)^d
Gender (female)	0.72 (0.31-1.66)	0.72 (0.29-2.39)	0.24 (0.05-1.181)	0.60 (0.19-1.82)
Etiology (HCV versus non-HCV)	0.69 (0.29-1.64)	0.86 (0.30-2.40)	0.39 (0.100-1.181)	1.07 (0.37-3.07)
SOFA (admission)	0.93 (0.85-1.03)	1.07 (0.94-1.22)
SOFA (48 hours)	0.88 (0.79-0.99)		1.04 (0.87-1.23)
SOFA (day of transplant)	0.94 (0.85-1.04)			1.04 (0.90-1.19)
χ² (degrees of freedom)		31.00 (9)	35.73 (8)	23.81 (8)

All three multivariable models were adjusted for site (five) and decade of transplant (2000 through 2009 versus 1990 through 1990). Hosmer-Lemeshow Goodness of Fit performed well (P > 0.15) for all models. Model one included age, gender (female), etiology (HCV versus non-HCV) and SOFA on admission (n = 180/198 patients; 18 patients had missing data). Model two had similar covariates but incorporated SOFA at 48 hours after admission (n = 110; 88 patients had missing data). Model three had similar covariates but incorporated SOFA on the day of liver transplant (n = 140; 58 patients had missing data). Significant P values were for age (unadjusted (P = 0.013),^a Model 1 (P = 0.015),^b Model 2 (P = 0.014), and ^d Model 3 (P = 0.027).

ISSN: 1364-8535