Skip to main content

Table 6 Results of the association parameter α obtained from joint model and TVCM fitted to data generated considering a non-monotonic baseline hazard function (scenario 9 of Table 1), α(0.3,0.6) and σε(0.1,0.3,0.5) with CV (3.1%,9.4%,15.6%)

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

  α=0.3α=0.6
σεModelEstESEASE%BiasCPEstESEASE%BiasCP
0.1TVCM(1x/week)0.2920.0580.058-2.7940.5790.0620.061-3.593
 joint-constant0.1830.0410.054-39390.3360.0350.052-440
 joint-weibull0.3370.0660.05812860.6380.0650.0616.389
 joint-spline0.3030.0600.0591.0950.6080.0640.0631.394
0.3TVCM(1x/week)0.2740.0560.056-8.7920.5380.0590.058-10.380
 joint-constant0.1800.0430.055-40390.3280.0370.053-450
 joint-weibull0.3400.0680.05913870.6420.0700.0657.089
 joint-spline0.3040.0620.0611.3950.6080.0770.0671.394
0.5TVCM(1x/week)0.2440.0530.053-18810.4710.0550.055-2136
 joint-constant0.1740.0460.052-42390.3120.040.055-480
 joint-weibull0.3440.0710.06215850.6430.1640.0717.289
 joint-spline0.3040.0650.0641.3950.6150.1970.0732.594
  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