Skip to main content

Table 3 Binomial regression model for signature inclusion of microarray feature BX281671 (ITGA2B) in different modeling approaches (sh: Fine-Gray regression; csh: cause-specific hazard model)

From: Dealing with prognostic signature instability: a strategy illustrated for cardiovascular events in patients with end-stage renal disease

Term ITGA2B   All features
  estimate p-value   sig - sig +
Intercept −3.82 <0.001   - -
sh vs. csh −0.16 0.013   24 16
Outlier excluded 0.58 <0.001   29 19
Original endpoint 2.63 <0.001   30 23
Univariate model −0.98 <0.001   31 23
sh × outlier excluded −0.01 0.798   0 0
sh × original endpoint −0.12 0.056   19 17
sh × univariate −1.27 <0.001   31 25
Outlier excluded × original endpoint −0.43 <0.001   19 11
Original endpoint × univariate 0.15 0.055   21 14
  1. The numbers of significant effects (5 % level after Bonferroni correction) for all 58 microarray features with inclusion frequency larger or equal to 0.1 in any of the approaches are indicated in the two rightmost columns, separately by positive and negative signs