Skip to main content

Table 4 Results of the association parameter α obtained from joint model and TVCM fitted to data generated considering the sample size n(35,75,150,300,600) and different probability distributions (scenarios 3:6 of Table 1) for the random effect bi0 with variance Σ11=1.44, α=0.3 and σε=0.3 with CV =9.4%

From: Joint model robustness compared with the time-varying covariate Cox model to evaluate the association between a longitudinal marker and a time-to-event endpoint

  Joint modelTVCM(1x/week)
DistributionnEstESEASE%BiasCPEstESEASE%BiasCP
 350.3150.1910.1795.0950.2870.1820.176-4.395
 750.3080.1230.1162.7940.2860.1170.114-4.795
Normal1500.3060.0850.0812.0940.2850.0810.079-5.094
 3000.3020.0570.0560.7950.2820.0550.055-6.094
 6000.3040.0400.0401.3950.2840.0380.039-5.393
 350.3230.0540.0517.7950.3150.0660.0625.096
 750.3090.0330.0333.0950.3030.0380.0381.095
Bimodal1500.3050.0230.0231.7950.3010.0250.0250.396
 3000.3020.0160.0160.7960.2990.0170.018-0.396
 6000.3020.0110.0110.7960.2990.0120.012-0.395
 350.3660.2560.21122930.3090.2150.1953.095
 750.3340.1340.12511940.2970.1210.118-1.096
Chisquare1500.3160.0880.0835.3940.2870.0790.079-4.395
 3000.3180.0590.0576.0940.2890.0540.054-3.795
 6000.3090.0400.0393.0950.2830.0360.037-5.794
 350.3520.2320.20017930.3050.1970.1871.796
 750.3270.1350.1229.0930.2910.1200.116-3.096
Gamma1500.3160.0790.0815.3960.2870.0730.077-4.396
 3000.3160.0610.0565.3940.2890.0560.054-3.794
 6000.3110.0410.0393.7940.2860.0370.037-4.795
  1. Mean of the maximum likelihood estimates (Est), empirical Monte Carlo standard error (ESE), asymptotic standard error (ASE), percentage bias (%Bias) and 95% coverage probabilities (CP) are shown