Application of Seemingly Unrelated Regression in Medical Data with Intermittently Observed Time-Dependent Covariates
Table 5
Simulation results (summary of estimated parameters, bias, and MSE) for intermittent observation of time-dependent covariates (situation 9 from Table 1) with different sample sizes (, and 240).
Parameter (true value)
Method
Sample size = 20
Sample size = 60
Sample size = 120
Sample size = 240
Bias
MSE
Bias
MSE
Bias
MSE
Bias
MSE
(−1.13)
SURCC
−.95
.18
6.64
−1.14
−.01
1.86
−1.12
.01
.83
−1.13
−.003
.43
MRMCC
−.99
.14
7.43
−1.16
−.03
2.007
−1.13
.002
.89
−1.14
−.01
.46
SURBOCF
−1.04
.09
7.90
−1.14
−.01
1.92
−1.12
.008
.83
−1.14
−.006
.42
MRMBOCF
−1.03
.10
6.53
−1.16
−.03
1.85
−1.12
.006
.82
−1.14
−.01
.42
SURLOCF
−1.04
.09
7.90
−1.14
−.01
1.92
−1.12
.008
.83
−1.14
−.006
.42
MRMLOCF
−1.03
.10
6.53
−1.16
−.03
1.85
−1.12
.006
.82
−1.14
−.01
.42
(.23)
SURCC
.20
−.03
6.77
.14
−.09
1.90
.20
−.02
.88
.24
.007
.44
MRMCC
.20
−.02
6.70
.12
−.11
2.008
.19
−.04
.93
.23
−.002
.45
SURBOCF
.24
.01
7.53
.14
−.08
1.93
.21
−.02
.87
.24
.01
.42
MRMBOCF
.17
−.05
5.95
.12
−.11
1.85
.19
−.03
.86
.24
.01
.43
SURLOCF
.24
.01
7.53
.14
−.08
1.93
.21
−.02
.87
.24
.01
.42
MRMLOCF
.17
−.05
5.95
.12
−.11
1.85
.19
−.03
.86
.24
.01
.43
(−.01)
SURCC
−.11
−.10
1.58
−.007
.003
.41
−.02
−.01
.20
−.02
−.005
.09
MRMCC
−.02
−.008
4.16
.01
.003
1.13
−.03
−.02
.56
−.02
−.02
.25
SURBOCF
.02
.03
4.10
.02
.02
.93
−.007
.002
.46
−.01
−.001
.20
MRMBOCF
−.03
−.02
3.38
.02
.03
.88
.002
.01
.44
.005
.01
.20
SURLOCF
.02
.03
4.10
.02
.02
.93
−.007
.002
.46
−.01
−.001
.20
MRMLOCF
−.03
−.02
3.38
.02
.03
.88
.002
.01
.44
.005
.01
.20
(−1.12)
SURCC
−.19
.93
2.02
−.28
.84
1.001
−.27
.85
.86
−.26
.86
.80
MRMCC
−1.14
−.02
2.89
−1.16
−.04
.77
−1.09
.03
.37
−1.10
.02
.18
SURBOCF
−1.05
.07
2.82
−1.02
−.09
.71
−.96
.15
.34
−.99
.13
.17
MRMBOCF
−.95
.16
2.27
−.97
.15
.65
−.92
.20
.35
−.94
.18
.18
SURLOCF
−1.05
.07
2.82
−1.02
−.09
.71
−.96
.15
.34
−.99
.13
.17
MRMLOCF
−.95
.16
2.27
−.97
.15
.65
−.92
.20
.35
−.94
.18
.18
*Note: , and MRMLOCF represent the estimated parameters from MRM and SUR models under complete case (CC) analysis, baseline observation carried forward (BOCF), and last observation carried forward (LOCF) method, respectively.