Table 4. Multivariate Cox-regression analysis with variable selection by AIC criterion.
Variable | Multivariate analysis | ||
HR | 95% CI | p-value | |
Sex (female) | 4.727 | 0.834–26.779 | 0.0792 |
WHO grade II | 3.239 | 0.894–11.733 | 0.0735 |
WHO grade III | 10.347 | 1.325–80.798 | 0.0259 |
TIMP3b | 0.175 | 0.029–1.043 | 0.0557 |
1p36 | 10.447 | 2.266–48.164 | 0.0026 |
22q | 0.411 | 0.135–1.251 | 0.1174 |
Results for multivariate Cox regression models with variable selection. For every variable, the estimated hazard ratio HR and its corresponding 95% confidence interval are given, as well as the p-value when testing HR = 1. The categorical variable WHO grade is treated as factor variable, thus HR refers to the comparison of WHO grade II and WHO grade III, respectively, to WHO grade I. The full model before variable selection included the additional variables age, p16, and 9p which were eliminated by stepwise backward selection with standard AIC (Akaike Information Criterion) as model selection criterion. All estimates and p-values in the table are close to the corresponding values before variable selection (see Table 3).