Skip to main content

Table 3 Estimated regression parameters from analyses of Kaiser data based on an illness-death model with B-spline parameterized baseline hazard functions

From: Fitting a shared frailty illness-death model to left-truncated semi-competing risks data to examine the impact of education level on incident dementia

 

With shared frailty

Without shared frailty

 

Years since age 65

Years since study entry

Years since age 65

Years since study entry

Parameter

HR

95% CI

p-value ∗

HR

95% CI

p-value ∗

HR

95% CI

p-value ∗

HR

95% CI

p-value ∗

Frailty variance, θ

0.41

(0.37,0.45)

<.001

0.25

(0.2,0.31)

<.001

      

Dementia

Trade degree, College

0.87

(0.83,0.92)

<.001

0.89

(0.85,0.94)

<.001

0.88

(0.84,0.92)

<.001

0.91

(0.87,0.95)

<.001

Post-graduate

0.76

(0.71,0.81)

 

0.80

(0.74,0.85)

 

0.81

(0.76,0.86)

 

0.83

(0.78,0.88)

 

Female

1.01

(0.96,1.05)

0.40

1.02

(0.98,1.07)

0.17

1.00

(0.95,1.04)

0.44

1.03

(0.99,1.08)

0.06

Black

1.42

(1.34,1.51)

<.001

1.42

(1.34,1.51)

<.001

1.31

(1.24,1.38)

<.001

1.38

(1.31,1.46)

<.001

Asian

0.81

(0.74,0.89)

 

0.90

(0.82,0.98)

 

0.80

(0.73,0.87)

 

0.90

(0.83,0.98)

 

Latinx

0.99

(0.89,1.09)

 

1.07

(0.97,1.17)

 

1.07

(0.98,1.16)

 

1.06

(0.97,1.16)

 

Age at study entry

   

2.03

(1.98,2.08)

<.001

   

1.97

(1.93,2.02)

<.001

Death

Trade degree, College

0.84

(0.81,0.88)

<.001

0.85

(0.81,0.88)

<.001

0.84

(0.81,0.87)

<.001

0.88

(0.85,0.92)

<.001

Post-graduate

0.68

(0.65,0.72)

 

0.69

(0.65,0.73)

 

0.73

(0.69,0.77)

 

0.75

(0.71,0.79)

 

Female

0.62

(0.60,0.65)

<.001

0.62

(0.60,0.64)

<.001

0.62

(0.60,0.65)

<.001

0.64

(0.62,0.67)

<.001

Black

1.02

(0.96,1.07)

<.001

1.05

(0.99,1.11)

<.001

1.03

(0.98,1.08)

<.001

1.01

(0.96,1.06)

<.001

Asian

0.70

(0.65,0.76)

 

0.77

(0.71,0.83)

 

0.81

(0.76,0.87)

 

0.76

(0.70,0.82)

 

Latinx

0.72

(0.65,0.79)

 

0.78

(0.71,0.85)

 

0.76

(0.70,0.82)

 

0.77

(0.71,0.84)

 

Age at study entry

   

1.62

(1.59,1.66)

<.001

   

1.57

(1.55,1.60)

<.001

Death following dementia

Trade degree, College

0.90

(0.85,0.96)

<.001

1.02

(0.96,1.09)

0.52

0.99

(0.93,1.05)

0.001

1.03

(0.97,1.10)

0.23

Post-graduate

0.77

(0.71,0.84)

 

1.05

(0.96,1.15)

 

0.86

(0.79,0.93)

 

1.07

(0.99,1.16)

 

Female

0.67

(0.63,0.70)

<.001

1.02

(0.95,1.08)

0.31

0.68

(0.64,0.71)

<.001

1.01

(0.95,1.07)

0.37

Black

0.86

(0.79,0.92)

<.001

1.02

(0.94,1.10)

0.19

0.95

(0.88,1.02)

<.001

0.99

(0.92,1.06)

0.90

Asian

0.69

(0.60,0.78)

 

1.09

(0.97,1.24)

 

0.69

(0.61,0.78)

 

1.01

(0.90,1.13)

 

Latinx

0.74

(0.64,0.84)

 

1.08

(0.95,1.23)

 

0.90

(0.80,1.02)

 

1.00

(0.89,1.13)

 

Age at dementia

   

1.01

(0.98,1.03)

0.31

   

1.00

(0.98,1.03)

0.37

  1. Models were fit with and without a shared frailty term. Two time origins were considered: 1) age 65 (age scale, left-truncated data); and 2) study entry (time-on-study)
  2. *p-value for a categorical variable with k>2 levels (educational attainment, race/ethnicity) is based on a (k−1)-degree of freedom Wald test of linear hypotheses
Back to article page

BMC Medical Research Methodology

ISSN: 1471-2288

Contact us