Skip to main content

Advertisement

Table 3 Predictive ability of the two models and of the phase-2 variable.

From: Multiple imputation for estimating hazard ratios and predictive abilities in case-cohort surveys

   Full cohort    Multiple imputation  
  Est SE SE % H 0 rejected Est SE SE % H 0 rejected
Z2 normally distributed         
β 1 = β 2 = β 3 = 0         
C 1 0.518 0.033 0.012   0.518 0.033 0.012  
C 2 0.524 0.033 0.013   0.526 0.033 0.013  
Δ 0.006 0.010 0.010 3.7 0.008 0.012 0.010 3.4
β 1 = β 2 = β 3 = log(2)         
C 1 0.733 0.029 0.014   0.733 0.029 0.014  
C 2 0.783 0.027 0.013   0.783 0.027 0.014  
Δ 0.049 0.010 0.010 100 0.049 0.010 0.011 100
Z2 normally distributed         
β 1 = β 2 = β 3 = 0         
C 1 0.518 0.033 0.012   0.518 0.033 0.012  
C 2 0.524 0.033 0.013   0.520 0.031 0.016  
Δ 0.006 0.010 0.009 5.5 0.002 0.013 0.012 4.2
β 1 = β 2 = β 3 = log(2)         
C 1 0.784 0.013 0.006   0.784 0.013 0.006  
C 2 0.881 0.011 0.006   0.866 0.011 0.006  
Δ 0.097 0.005 0.005 100 0.082 0.005 0.004 100
Z2 uniformly distributed         
β 1 = β 2 = β 3 = 0         
C 1 0.532 0.055 0.019   0.532 0.055 0.019  
C 2 0.540 0.055 0.019   0.541 0.055 0.020  
Δ 0.008 0.015 0.013 2.2 0.009 0.017 0.013 4.0
β 1 = β 2 = β 3 = log(2)         
C 1 0.733 0.029 0.014   0.733 0.029 0.014  
C 2 0.781 0.027 0.012   0.785 0.027 0.012  
Δ 0.048 0.009 0.009 100 0.052 0.010 0.010 100
  1. Results of 1000 simulations
  2. C 1, Harrell's C index of the proportional hazard model without the phase-2 variable
  3. C 2, Harrell's C index of the proportional hazard model with the phase-2 variable
  4. Δ, Harrell's predictive value of the phase-2 variable, H0: Δ = 0