Table 4.

Differences between chimney sweeps and control group in serum markers of cardiovascular disease, explored with a general-linear-model. Effect estimates are presented as β-values [95%CI = 95% confidence interval].

Variable	Model 1f		Model 2g		Model 3h
	P	β1 (CI 95%)	P	β1 (CI 95%)	P	β1 (CI 95%)
CRPa	0.709	−0.16 (−0.98, 0.67)	0.938	0.04 (−0.98, 1.06)	0.747	−0.15 (−1.04, 0.75)
GGTb	0.193	0.06 (−0.03, 0.16)	0.497	0.04 (−0.07, 0.14)	0.201	0.07 (−0.04, 0.17)
Homocysteine	<0.001	3.37 (1.67, 5.07)	<0.001	2.86 (1.36, 4.37)	0.001	2.96 (1.18, 4.74)
HDLc	0.004	0.13 (0.04, 0.22)	0.011	0.13 (0.03, 0.23)	0.001	0.16 (0.06, 0.26)
Cholesterol	0.003	0.43 (0.15, 0.72)	0.015	0.39 (0.08, 0.71)	0.005	0.44 (0.13, 0.75)
LDLd	0.061	0.25 (−0.01, 0.51)	0.225	0.18 (−0.11, 0.47)	0.070	0.26 (−0.02, 0.55)
TGe	0.190	0.19 (−0.10, 0.48)	0.074	0.26 (−0.03, 0.54)	0.446	0.12 (−0.19, 0.43)

^aC-reactive protein, ^bgamma-glutamyl transferase, ^chigh-density lipoprotein, ^dlow-density lipoprotein, ^etriglycerides. ^f(serum marker) = intercept + β1 × exposure group (2 categories) + β2 × age (continuous) + β3 × BMI (continuous) + β4 × smoking status (4 categories) + e (residual error). ^g(serum marker) = intercept + β1 × exposure group (2 categories) + β2 × age (continuous) + β3 × BMI (continuous) + e (residual error). Current and party smokers were excluded from this analysis. ^h(serum marker) = intercept + β1 × exposure group (2 categories) + β2 × age (continuous) + β3 × BMI (continuous) + β4 × smoking status (4 categories) + β5 × use of snus (2 categories) + β6 × physical activity (4 categories) + β7 × passive smoking (2 categories) + β8 × current residence (4 categories) + β9 × education (2 categories) + β10 × family history of cardiovascular disease (2 categories) + β11 × exposure to polycyclic aromatic hydrocarbons from hobby (2 categories) + e (residual error).