Table 3.

Internal dose of cadmium at follow-up predicted from characteristics at baseline (1985–1989).

Variable	Log BCd (nmol/L)	Log BCd (nmol/L)	Log UCd (nmol/24 hr)
Follow-up period	1991–1996	2001–2003	1991–1996
No. of subjects analyzed	711	461	730
Median follow-up [years (IQR)]	5.2 (4.8–5.5)	13.4 (10.7–16.0)	5.2 (4.8–5.5)
R2	0.61	0.58	0.63
Intercept (β± SE)	0.229 ± 0.090#	0.448 ± 0.093##	0.201 ± 0.075#
Partial regression coefficients (β± SE) Being female (0,1)	−0.0068 ± 0.0188	0.0051 ± 0.0198	−0.0300 ± 0.0160
Baseline (1985–1986) value (β± SE)
Log BCd (nmol/L)	0.681 ± 0.033##	0.540 ± 0.034##	—
Log UCd (nmol/day)	—	—	0.664 ± 0.028##
Age (+10 years)	0.052 ± 0.007##	0.044 ± 0.007##	0.031 ± 0.006##
Body mass index (+1 kg/m2)	−0.0009 ± 0.0020	−0.0016 ± 0.0023	0.0019 ± 0.0017
Smoking (0,1)	0.086 ± 0.021##	0.084 ± 0.022##	0.073 ± 0.015##
Log serum ferritin (pmol/L)	−0.110 ± 0.024##	−0.111 ± 0.025**	−0.040 ± 0.020*
Log distance to nearest smelter (km)	−0.144 ± 0.016##	−0.069 ± 0.016##	−0.034 ± 0.014**

—, covariable not considered in the analysis. The dependent variable is the logarithmically transformed variable reflecting the internal cadmium dose at follow-up.

Significance of the partial regression coefficients:

^*0.10 ≤ p < 0.05;

^**p ≤ 0.05;

^#p ≤ 0.01;

^##p ≤ 0.001. Being female (0, 1) and body mass index were not selected by stepwise regression (p ≥ 0.22), but were forced into the models.