[Retracted] A Joint Model for Unbalanced Nested Repeated Measures with Informative Drop-Out Applied to Ambulatory Blood Pressure Monitoring Data
Table 3
Parameter estimates for DBP and informative dropout association based on the joint model and two-stage approach.
Parameter estimate
Joint model
Two-stage model∧∧
Longitudinal submodel for DBP∧
Mean
SD
95% CI
Mean
SD
95% CI
White race ()
2.9155
0.0919
(2.6600, 3.0470)
4.0091
0.0100
(3.9800, 4.0300)
BMI -score ()
0.7068
0.0067
(0.5653, 0.8165)
0.0086
0.0000
(0.000, 0.017)
Variation components, random effects
Between subjects, intercepts
5.6011
0.6043
(4.5209, 6.8862)
0.0600
Covariance, intercept, and visits
−0.1881
0.0405
(−0.2781, −0.1170)
Covariance, intercept, and slopes
0.0112
0.0167
(−0.0214, 0.0439)
Between subjects, visits
0.0246
0.0046
(0.0171,0 .0353)
0.0400
Covariance, visits, and slopes
−0.0006
0.0011
(−0.0027, 0.0014)
Between subjects, slopes
0.0058
0.0006
(0.0047, 0.0071)
0.0400
Measurement error
0.0159
0.0002
(0.0156, 0.0162)
0.1200
Weibull submodel for dropout
Shape ()
0.7388
0.1035
(0.5456, 0.9610)
1.0000
1.0300
(0.9600, 1.0400)
Intercept ()
4.6567
5.2972
(−6.2182, 14.0500)
Age ()
−0.0365
0.0609
(−0.1572, 0.0818)
−0.0000
0.03900
(−0.2900, 0.4400)
Male gender ()
−0.2594
0.2658
(−0.7776, 0.2553)
0.0680
0.1870
(0.0680, −0.2500)
BMI -score ()
−1.7578
0.9144
(−3.3892, 0.1121)
0.0800
0.0800
(−0.7600, 0.2500)
White race ()
−8.2174
3.8876
(−15.5100, −0.2371)
−0.2400
0.1900
(−0.6200, 0.1300)
Mild group ()
−0.4286
0.3381
(−1.1130, 0.2078)
0.1500
0.2300
(−0.3000, 0.6000)
Severe group ()
−0.1986
0.3650
(−0.9393, 0.5088)
0.2400
0.2300
(−0.2200, 0.6900)
Random effects
Random intercepts ()
−2.7811
1.3194
(−5.2860, −0.0579)
−1.4000
1.9700
(−5.2700, 2.4700)
Random visits ()
−22.9998
3.6157
(−30.3305, −16.1298)
−1.1700
3.2700
(−7.5700, 5.2300)
Random slopes ()
−1.3591
9.8496
(−20.4918, 18.0111)
−19.5000
4.3200
(−27.9700, −11.0200)
BMI: body mass index; CI: credible interval; DBP: diastolic blood pressure; SD: standard deviation of parameter estimate. ∧Group-specific splines included in the longitudinal submodel are reported in the supplement. Tables S6 and S7 in Section S4: the coefficients of the transformed natural cubic spline functions for the two-stage approach and the joint model; ∧∧parameter estimates evaluated using Wald test; 95% CI excludes 0.
