Joint modeling of multivariate longitudinal data and the dropout process in a competing risk setting: application to ICU data

Table 5 Simulation study

	Positive association		Negative association
Parameter	True value	Bias (SD)	True value	Bias (SD)
n = 100
Longitudinal
Intercept β ₁₀	6.15	0.24 (0.16)	6.15	0.20 (0.16)
Time β ₁₁	-0.25	0.22 (0.03)	-0.25	0.19 (0.04)
Binary covariate β ₁₂	0.25	0.01 (0.19)	0.25	0.01 (0.20)
Interaction Time × Binary covariate β ₁₃	0.0	-0.03 (0.03)	0.0	-0.05 (0.03)
Survival
Binary covariate β ₂	0.0	0.25 (0.34)	0.0	0.39 (0.35)
γ	1.0	-0.09 (0.12)	-1.0	0.13 (0.12)
Variances
σ_U0	1.0	0.04 (0.14)	1.0	0.02 (0.13)
σ_U1	1.0	0.02 (0.13)	1.0	0.02 (0.13)
n = 500
Longitudinal
Intercept β ₁₀	6.15	0.20 (0.16)	6.15	0.23 (0.04)
Time β ₁₁	-0.25	0.02 (0.02)	-0.25	0.05 (0.05)
Binary covariate β ₁₂	0.25	0.04 (0.22)	0.25	0.03 (0.06)
Interaction Time × Binary covariate β ₁₃	0.0	-0.007 (0.04)	0.0	-0.01 (0.01)
Survival
Binary covariate β ₂	0.0	0.12 (0.29)	0.0	-0.16 (0.12)
γ	1.0	-0.03 (0.09)	-1.0	-0.04 (0.10)
Variances
σ_U0	1.0	0.02 (0.14)	1.0	0.04 (0.12)
σ_U1	1.0	0.03 (0.14)	1.0	0.03 (0.13)

Simulation study, bias and standard deviations (SD) estimates of the posterior means for competing risk joint model, sample size = 100, 500.

ISSN: 1471-2288