Skip to main content

Table 11 Mixed linear regression for log RR with a random intercept with weights at the first level (exposure categories level) proportional to the inverse of the respective variance and with the totals of these weights as study weights at the second level (study level)

From: Diesel engine exhaust and lung cancer risks – evaluation of the meta-analysis by Vermeulen et al. 2014

log RR

Coef.

Robust Std. Err.

Z

P

95 % CI

Dose

0.00097

0.00014

6.74

<0.001

0.00069

0.00125

Constant

0.084

0.111

0.75

0.45

−0.134

0.300

Garshick (modified)

Dose

0.00087

0.00019

4.58

<0.001

0.00050

0.00124

Constant

0.024

0.062

0.39

0.70

−0.097

0.145

Crump et al. [14], REC4

Dose

0.00045

0.00021

2.10

0.035

0.00003

0.00086

Constant

0.178

0.094

1.90

0.057

−0.005

0.362

Crump et al. [14], REC4 and Garshick (modified)

Dose

0.00048

0.00023

2.10

0.036

0.00003

0.00092

Constant

0.097

0.040

2.39

0.017

0.018

0.176

Crump et al. [14], REC6

Dose

0.00053

0.00026

2.05

0.040

0.00002

0.00103

Constant

0.172

0.101

1.71

0.088

−0.025

0.369

Crump et al. [14], REC6 and Garshick (modified)

Dose

0.00052

0.00025

2.09

0.036

0.00003

0.00101

Constant

0.095

0.051

1.87

0.061

−0.044

0.194

  1. The weights of the first level are scaled effectively to the weights of the second level. Coefficient (Coef.) of the cumulative exposure (Dose) to DEE in μg/m3-years and absolute term (Constant) with standard deviation (Std.Err.); Z-test statistic for the coefficient with the corresponding P-value and a 95 % confidence interval for the coefficient, 95 % CI. Silverman et al. [4] was incorporated as used by Vermeulen et al. [1], and in accordance to the re-analysis by Crump et al. [14] with adjustment for the radon exposure (REC4, REC6, cf. Table 3). Garshick et al. [3] as used by Vermeulen et al. [1], and Garshick et al. [3] (modified) were incorporated