Table 2 Comparing coefficients of full entry, stepwise and LASSO Cox Regression models for risk prediction of silicosis.

From: Prediction models and risk assessment for silicosis using a retrospective cohort study among workers exposed to silica in China

Regression models

Model1

Model2

Model3

Age at entry of the cohort (years)

 <20 y

ref.

ref.

ref.

 20 ~ 24 y

0.97*

0.97*

0.69

 25–29 y

1.43**

1.43**

1.17

 30 ~ 34 y

1.92***

1.92***

1.66

 >=35 y

2.13***

2.13***

1.87

 Test for trend

0.44***

0.44***

0.43

Mean concentration of respirable silica dust exposure (mg/m3)

 <0.05

ref.

ref.

ref.

 0.05 ~ 0.10

1.43***

1.43***

1.31

 0.10 ~ 0.15

2.51***

2.51***

2.38

 >=0.15

3.26***

3.26***

3.13

 Test for trend

1.00***

1.00***

1.00

Net years of dust exposure (years)

 

 <10

ref.

ref.

ref.

 10 ~ 15

1.33***

1.33***

1.24

 15 ~ 20

1.42***

1.42***

1.33

 20 ~ 25

1.65***

1.65***

1.56

 >=25

1.04***

1.04***

0.94

 Test for trend

0.12***

0.12***

0.12

No. of job experienced in the iron ore

 1 job

ref.

ref.

ref.

 2 jobs

0.61***

0.61***

0.58

 3 jobs or more

0.62***

0.62***

0.60

 Test for trend

0.34***

0.34***

0.33

Smoking habits

 Non-smoker

ref.

ref.

ref.

 Smoker

0.56***

0.54***

0.53

 Test for trend

0.58***

0.58***

0.57

Illiteracy a

 No

ref.

ref.

ref.

 Yes

0.38**

0.38**

0.38

 Test for trend

0.37**

0.37**

0.37

  1. Abbreviation: Model1, Cox model with all predictors entered at the same time; Model2, Cox model with stepwise selection (selection entry p = 0.05, removing p = 0.051); Model3, Cox model with LASSO selection;
  2. *p value < 0.05; ** p value < 0.01; *** p value < 0.0001.
  3. aWorkers whose education attainment were below primary school;
Back to article page