Table 5.
Multivariable analysis of survival in all patients
| HR (95% CI) | P-value | |
|---|---|---|
| Disease (NAFLD) | 1.22 (0.97–1.54) | 0.084 |
| Cirrhotic | 1.91 (1.18–3.08) | 0.008* |
| Largest HCC Size (cm) | <0.001* | |
| <2.0 | ||
| 2.0-4.9 | 1.84 (1.28–2.63) | <0.001* |
| 5.0+ | 2.68 (1.77–4.04) | <0.001* |
| AFP | <0.001* | |
| <5 | ||
| 5–24 | 1.34 (0.98–1.83) | 0.068 |
| 25–249 | 1.80 (1.32–2.47) | <0.001* |
| 250+ | 2.14 (1.53–3.01) | <0.001* |
| Vascular invasion | 1.79 (1.26–2.53) | 0.001* |
| Radiofrequency ablation | 0.69 (0.50–0.96) | 0.026* |
| Liver resection | 0.51 (0.27–0.97) | 0.041* |
| Sorafenib therapy | 0.31 (0.20–0.48) | <0.001* |
| Transplanteda | 0.55 (0.37–0.82) | 0.003* |
Results from a multivariable cox regression model, using a forward stepwise entry procedure. The disease group was forced into the model, and factors in Tables 1–3 were considered for inclusion as additional confounders. The initial model did not identify any of the factors with >10% missing data as significant predictors of survival, hence these were excluded to maximize the available sample size, and the model re-run. The final model was based on the n = 355 cases with data available for all of the factors considered.
HR, hazard ratio.
aTreated as a time-dependent covariate, in order to account for the effect of survivor bias.
*
Significant at P < 0.05.