requires more research. Several biological mechanisms linking air pollution to atherosclerosis and increased risk for cardiovascular disease has
been suggested, including endothelial dysfunction, prothrombotic and
coagulant changes, systemic inflammation and oxidative stress; many
of these triggered by exposure to particles (Brook et al., 2010). The present study indicates that effects of air pollution on levels of blood cholesterol, a known risk factor for cardiovascular morbidity and mortality
(Pencina et al., 2009; Prospective Studies Collaboration et al., 2007), is
another possible biological mechanism. A previous study found a
39 mg/dl lower total cholesterol to be associated with a 34% lower mortality for ischemic heart disease in the age group 50–69 years
(Prospective Studies Collaboration et al., 2007), corresponding to the
age group of the participants in the present study. We find a 1.0 μg/m3
higher exposure to PM2.5 to be associated with a 0.7 mg/dl higher cholesterol in the total study population, and 4.7 mg/dl higher cholesterol
among diabetics. Although the observed differences in cholesterol are
small, they may have an impact on public health, considering the widespread nature of air pollution.
Interestingly, we found an interaction between the two air pollutants, with a higher association between NO2 and cholesterol among participants exposed to high levels of PM2.5 as compared to participants
exposed to low levels of PM2.5. Also, both PM2.5 and NO2 seemed associated with cholesterol in models including both air pollutants, indicating
independent effects. Exposure to particles have for many years been
thought important in relation to the development of cardiovascular disease (Brook et al., 2010). NO2 is an indicator for vehicles emission and
traffic, and, thus, a marker of a number of pollutants, including ultrafine
particles.
The present study suggested that road traffic noise may be associated with slightly higher cholesterol levels, though results were only
borderline significant. Two studies have investigated associations between road traffic noise and cholesterol (without adjustment for air pollution), and while one indicated a positive association with the level of
cholesterol (Babisch et al., 1988), the other found no association
(Babisch et al., 1993). Biological mechanisms behind a potential association between noise and cholesterol levels are not clear. Exposure to
road traffic noise is known to result in sleep disturbance, and studies
have indicated that short and disturbed sleep is associated with higher
levels of cholesterol (Ekstedt et al., 2004; Gangwisch et al., 2010; Wan
Mahmood et al., 2013). While air pollution, thus, might affect levels of
cholesterol through a relatively direct exposure to particles and other
air pollutants, road traffic noise may affect cholesterol levels more indirectly, through disturbance of sleep.
The association between noise and cholesterol weakened after adjustment for PM2.5 and disappeared after adjustment for NO2. Road traffic noise and air pollution are correlated in the present study reflecting
that road traffic is a source of both exposures, and therefore input variable for both noise and air pollution models. Adjustment for PM2.5 affected the noise-cholesterol estimates less than adjustment for NO2,
most likely reflecting the lower correlation between noise and PM2.5
as compared with noise and NO2. This illustrates that the relatively
high collinearity between the exposures (noise and air pollution) complicates the interpretation of results of the multiple pollutant models. A
main reason is that we cannot rule out that the air pollution models predicts air pollution levels more precisely than the noise model predicts
road traffic noise, which could potentially explain the apparent more robust association with air pollution. One study on the biological mechanisms behind the observed association between exposure and
cardiovascular disease has included both air pollution and road traffic
noise (Kalsch et al., 2014). They found that in mutually adjusted models
both long-term exposure to air pollution and nighttime traffic noise was
associated with subclinical atherosclerosis, estimated as thoracic aortic
calcification. However, with regard to road traffic noise they found no
significant association for Lden, suggesting that nighttime noise exposure is most hazardous. This suggests that Lden is not the optimal estimate for exposure to road traffic noise in the present study. However,
as exposure to road traffic noise during the night (Ln) was highly correlated with daytime exposure, Ld (Rs = 0.997) we could not separate the
effect of the two exposures. There is, thus, a need for studies with more
detailed information on nighttime noise exposure in order to investigate whether noise exposure is associated with blood cholesterol.
Also, studies on noise exposures relatively independent of air pollution,
such as airport noise, could be a valuable input in investigating whether
noise is an independent predictor of cholesterol levels as well as other
cardiovascular risk factors. Unfortunately, the present study population
include only 0.2% exposed to airport noise and, thus, too little power to
investigate airport noise in association with cholesterol levels.
We found that among participants with cardiovascular disease or diabetes, road traffic noise was associated with higher levels of cholesterol, suggesting that they are more susceptible. However, results
were insignificant and might be chance findings. Also, we found that
among diabetics, PM2.5 was more strongly associated with cholesterol
than among people without diabetes, again suggesting that diabetics
are more susceptible to traffic pollutants. Given the large increase in diabetes prevalence during the last decades, more research in this area
would be relevant.
Strengths of the present study included the large study population,
with detailed information on various potential confounders. Furthermore, access to residential address histories, state of the art exposure
models and high quality of input data to the exposure models improved
estimation of long-term exposure to road traffic noise and air pollution.
In addition, inclusion of both road traffic noise as well as two air pollutants estimated by two different models allowed for investigation of mutual confounding of the three correlated exposures.
A main limitation is the cross-sectional design, which prevents us
from making firm conclusions on causality and chronological order of
events, and results should, thus, be confirmed using a longitudinal design with repeated measures. Also, by deleting participants receiving
cholesterol-lowering medication we might have excluded the people
most susceptible to an effect of air pollution and noise on cholesterol,
and, thus, underestimated the association. However, less than 2% of
our population was excluded due to this, indicating that it was not a
major problem in the present study. Given today's excessive use of
cholesterol-lowering medication, our finding of only 2% receivers of
these drugs indicate an incomplete assessment. However, this study
was conducted between 1993 and 1997, where use of cholesterollowering medication was much less frequent in Denmark, and the percentage found in this study population did not deviate substantially
from the use in the general population in that time period (Riahi et al.,
2001). Another limitation is that we only had information on total cholesterol in blood, and not more specific cholesterol measures such as
low-density lipoprotein (LDL) cholesterol and high-density lipoprotein
(HDL) cholesterol. Estimation of air pollution and road traffic noise
was based on models, and although the AirGIS dispersion model, the
land-use regression model and the Nordic prediction method are all
standard methods used for many years, estimation of the exposures is
inevitably associated with some degree of uncertainty. Such misclassification is, however, likely to be non-differential, and influence the estimates towards the neutral value. We also lacked information on
individual factors that might influence exposure to noise or air pollution, such as information on time spend at home, workplace addresses,
bedroom location, indoor exposures etc., which might result in exposure misclassification and probably underestimation of the association.
A study on road traffic noise and myocardial infarction found that the
risk was underestimated when these factors was not considered
(Selander et al., 2009).