Table 3.

Linear mixed effect results describing the relationships between log-transformed wood dust levels and fixed effects. Results are based in 12653 measurements collected between 1978 and 2007.

Parameter	β	e	P-value	GMR	95% CI
Fixed effects
Intercept	1.294	0.218	<0.001	4.35	3.57 to 5.31
Year (ref 1997)	−0.081	0.002	<0.001	0.92	0.92 to 0.93
Sampling duration, min	−0.002	0.0001	<0.001	0.89*	0.88 to 0.90*
OAsJEM score
No exposure	−0.033	0.133	0.8	0.97	0.75 to 1.26
Low exposure	0.037	0.134	0.8	1.04	0.80 to 1.35
High exposure	Ref			Ref
Random effects
Country#
DK + NO	−0.733	0.1925	<0.001	0.48	0.33 to 0.70
NL	0.0488	0.1980	0.8	1.05	0.71 to 1.54
FI	0.1445	0.1949	0.5	1.16	0.79 to 1.69
United Kingdom	0.3623	0.1954	0.06	1.43	0.98 to 2.11
FR	0.1771	0.1922	0.3	1.19	0.82 to 1.74
Between-country variance (naive estimate)	0.182 (0.301)	0.13	0.08
Between-ISCO variance (naive estimate)	0.045 (0.109)	0.019	<0.01
Residual variance (naïve estimate^)	1.04 (1.09)	0.013	<0.0001
% of explained variance by the model
Between-country variance	39.4
Between-ISCO variance	58.6
Residual variance	14.9
Total variance	22.3

β = beta for log-transformed exposure levels, e = standard error, GMR = geometric mean ratio; 95% CI = confidence intervals for the estimated GMR, ISCO = International Standard Classification of Occupations 1988 edition (ISCO-88).

^*For an increase of 60 min in sampling time.

^{^}naïve estimates are derived from a model without fixed factors.

^#Entered as a random effect in the models, BLUP estimates shown.