Skip to main content

Table 2 Hazard ratios along with their 95% confidence intervals for the 12 most influential variables for the Cox models. Variables are presented in decreasing order according to the absolute z-score values (12.90 to 5.16) for the Cox model with all variables. Predictors shown are the most prognostic as their z-scores values correspond to low and very significant p-values. These variables were also selected by both Cox backward and Cox LASSO model which verifies their prognostic ability for Cox models

From: Survival prediction models since liver transplantation - comparisons between Cox models and machine learning techniques

 

Cox all variables

Cox backward

Cox LASSO

 

HR (95% CI)

HR (95% CI)

HR

Re-transplantation

1.602 (1.491-1.721)

1.608 (1.501-1.722)

1.558

Donor age

1.010 (1.008-1.011)

1.011 (1.009-1.012)

1.009

Donor type DCD (a)

1.483 (1.362-1.616)

1.443 (1.338-1.556)

1.298

log(Total cold ischemic time)

1.258 (1.192-1.327)

1.285 (1.221-1.353)

1.191

Diabetes

1.173 (1.125-1.225)

1.176 (1.128-1.226)

1.136

Race Black (b)

1.240 (1.171, 1.314)

1.261 (1.193-1.332)

1.186

Life support

1.343 (1.240-1.454)

1.375 (1.272-1.487)

1.304

Recipient age

1.007 (1.005-1.009)

1.008 (1.006-1.010)

1.006

Incidental tumour

1.314 (1.202, 1.437)

1.315 (1.203-1.437)

1.203

Hypertensive bleeding

1.296 (1.185, 1.418)

1.301 (1.190-1.423)

1.214

HCV (c) serology status

1.147 (1.091-1.206)

1.148 (1.094-1.205)

1.166

Pre-treatment status ICU (d)

1.240 (1.143, 1.346)

1.253 (1.160-1.354)

1.164

  1. (a): Donor type DCD (Donor Circulatory Dead) vs DBD (Donor after Brain-Dead), (b): Race Black vs White, (c): Chronic hepatitis C virus, (d): Intense Care Unit vs Non-hospitalised/Hospitalised