2.4. Statistical analyses
The present study is a cross-sectional study and we used generalized
linear models for the analyses. For each person, exposure to road traffic
noise and both air pollutants were modeled at baseline address as well
as a time-weighted 5-year means preceding enrol ment date for noise
and NO2, taking the complete migration history in this period into account.
We investigated the association between the three exposures
and level of total cholesterol at enrolment, including exposures as continuous
variables (per interquartile range). We adjusted for potential