Skip to main content
. 2019 Feb 11;16(3):504. doi: 10.3390/ijerph16030504

Table A2.

Parameter estimates for the within- between-subject decomposition regression of Table 2 EGFR = HIV infection, albumin and BUN in the Bronx WIHS using mixed models.

Variable Compartment Working Correlation Structure
Independence Equicorrelation
Point Estimate 95% CI 1 Z-Value (p) 1 Point Estimate 95% CI 1 Z-Value (p) 1
HIV Infection Between-subject
HIV,BS)
−1.16 (−4.21, 1.88) −2.30
(0.02)
−1.57 (−4.47, 1.32) −1.06
(0.29)
NA 2 --- --- --- NA 2 --- ---
Albumin
Per g/dL
Between-subject
ALB,BS)
−3.28 (−4.78,
−1.77)
−4.26
(0.16)
−2.72 (−7.00,
1.57)
−1.32 (0.19)
Within-subject
ALB,WS)
−10.70 (−12.24,
−9.15)
13.60 (<0.0001) −10.67 (−11.86,
−9.48)
−17.57 (<0.0001)
BUN Per mg/dL Between-subject
BUN,BS)
−2.72 (−2.84,
−2.60)
−44.87
(<0.0001)
−2.65 (−2.95,
−2.35)
−17.10
(<0.0001)
Within-subject
BUN,WS)
−1.11 (−1.22,
−0.99)
−18.59 (<0.0001) −1.11 (−1.20,
−1.03)
−25.72 (<0.0001)
Akaike Information Criteria (AIC) 98,451.8 94,824.7

1 The confidence interval and p-values for independence working correlation structure in particular but also arguably for equicorrelation overestimate the precision of the parameter estimates. Unlike GEE, mixed models are not robust to misspecification of the correlation structure. 2 There is no Within-subject Variation for HIV Infection Status.