Noise Health Home 

ARTICLE
[Download PDF]
Year : 2011  |  Volume : 13  |  Issue : 52  |  Page : 234--250

Cardiovascular effects of environmental noise: Research in Austria

Peter Lercher1, Dick Botteldooren2, Ulrich Widmann3, Ulrich Uhrner4, Ewald Kammeringer5,  
1 Department of Hygiene, Microbiology, and Social Medicine, Medical University Innsbruck Med Univ Innsbruck, Sonnenburgstra�e 16, A-6020 Innsbruck, Austria
2 Acoustics Group, Department of Information Technology, Gent University, Sint-Pietersnieuwstraat 41, B-9000 Gent, Belgium
3 AUDI AG, Abt. I/EK-5, D-85045 Ingolstadt, Germany
4 Technical University of Graz, Infeldgasse 21a, A-8010 Graz, Austria
5 University of Innsbruck, Technikerstrasse 13, A-6020 Innsbruck, Austria

Correspondence Address:
Peter Lercher
Department of Hygiene, Microbiology, and Social Medicine, Medical University Innsbruck, Sonnenburgstra�e 16, A-6020 Innsbruck
Austria

Abstract

Cardiovascular effects of noise rank second in terms of disability-adjusted life year (DALYs) after annoyance. Although research during the past decade has consolidated the available data base, the most recent meta-analysis still shows wide confidence intervals - indicating imprecise information for public health risk assessment. The alpine area of Tyrol in the Austrian part of the Alps has experienced a massive increase in car and heavy goods traffic (road and rail) during the last 35 years. Over the past 25 years small-, middle-, and large-sized epidemiological health surveys have been conducted - mostly within the framework of environmental health impact assessments. By design, these studies have emphasized a contextually driven environmental stress perspective, where the adverse health effects on account of noise are studied in a broader framework of environmental health, susceptibility, and coping. Furthermore, innovative exposure assessment strategies have been implemented. This article reviews the existing knowledge from these studies over time, and presents the exposure-response curves, with and without interaction assessment, based on standardized re-analyses and discusses it in the light of past and current cardiovascular noise effects research. The findings support relevant moderation by age, gender, and family history in nearly all studies and suggest a strong need for consideration of non-linearity in the exposure-response analyses. On the other hand, air pollution has not played a relevant role as a moderator in the noise-hypertension or the noise-angina pectoris relationship. Finally, different noise modeling procedures can introduce variations in the exposure response curves, with substantive consequences for public health risk assessment of noise exposure.



How to cite this article:
Lercher P, Botteldooren D, Widmann U, Uhrner U, Kammeringer E. Cardiovascular effects of environmental noise: Research in Austria.Noise Health 2011;13:234-250


How to cite this URL:
Lercher P, Botteldooren D, Widmann U, Uhrner U, Kammeringer E. Cardiovascular effects of environmental noise: Research in Austria. Noise Health [serial online] 2011 [cited 2023 Dec 4 ];13:234-250
Available from: https://www.noiseandhealth.org/text.asp?2011/13/52/234/80160


Full Text

 Introduction



Through its geographical position in central Europe, Austria has experienced transit-traffic since Roman times. Since the early seventies the Austrian part of the Alps has experienced a massive increase in car and especially in heavy goods traffic (road and rail). Currently, about 32% of the population is exposed to road noise levels ≥ 60 dBA and 60% ≥ 55 dBA. Even as the increase occurred foremost on the road, the first complaints were issued about highway traffic by the end of the seventies. In 1984, we started a first pilot study in a small community ('noise village'), which was surrounded by a highway and an associated toll station, to explore the problem. [1],[2] Later, the intentions to move heavy goods transport from road to rail led to an increase of heavy rail traffic during the night, resulting in higher noise levels than through the daytime (+3 dBA). Therefore, multi-community health surveys followed, to study the supposed adverse effects of noise and air pollution in those Alpine valleys where the transit-traffic was on the increase or where large rail infrastructure projects required environmental health impact assessment (EHIA). [3],[4],[5],[6],[7] In these studies we emphasized a contextually driven environmental stress perspective [8] and placed the study of noise-related adverse health effects in a broader framework of environmental health, susceptibility, and coping. [5],[9],[10],[11]

Cardiovascular effects of noise rank second in terms of disability adjusted life years (DALYs) after annoyance. Although research during the past decade has consolidated the available data base the most recent meta-analysis - based on road traffic noise studies - reveals wide confidence intervals. It is also not sufficiently clear what the inconsistent results, concerning standard potential effect modifiers such as sex, age, and education, mean. Furthermore, the quantitative role of the psychological and physiological vulnerability factors that can promote adverse effects of noise, such as, noise sensitivity and health status or family history of hypertension has not yet been fully understood. Even the strong data base on the cardiovascular effects of aircraft noise shows a substantial diversity in terms of exposure response shapes and slopes and in terms of observed effect modifiers. The conclusions about the effects of road traffic noise rest mainly on the Caerphilly and Speedwell and Berlin studies. Insufficient data are available on the potential cardiovascular effects on account of railway noise exposure.

In earlier articles we have suggested that the large variability of noise effects observed, is partly due to the strong moderation and / or mediation by the context where the noise exposure occurs and partly due to the effectiveness of the coping strategies. [9],[11],[12],[13] Related to this argument, factors such as regional differences in the underlying population morbidity structure (susceptibility and health status) and the overall exposure load (at work, environment, socioeconomic) may in addition be responsible for the often observed heterogeneous results. A specific argument is related to the potential difference in the experienced noise exposure, in the Alpine areas. This may be either related to the perception of noise (perceived exposure contrast, signal-to-noise ratio) or to the inability of classical noise indicators to catch the difference in the meaning of noise exposure, which is known to modify bodily responses. Eventually, since longitudinal studies are sparse and difficult to conduct in a continuously changing world with high mobility, the required latency time for the development of noise-associated cardiovascular effects has not yet been established. Thus, the sampled population experience in the studies may differ in terms of cumulative time to the effect, and reflect only the different power to detect effects, apart from the power provided by the sample size.

This article aims to share and integrate the existing knowledge from the Tyrol studies with a wider audience. The first step would be to make the analysis available, which has not yet been published - or if so - not in English. Second, to summarize the main results observed over a period of 25 years. Third, to add re-analysis based on the existing datasets, which would contribute to some of the still pertinent questions in cardiovascular noise effects research. For this purpose updated models were created to further evaluate the interaction effects and to gain a deeper insight into the meaning of effect modifiers over time.

 Methods



Area, sample selection, and recruitment

Both the areas of investigation, the Unterinntal and Wipptal, are located along the most important European North-South-access route for heavy goods, over the Brenner Pass. The heavy goods traffic over the Brenner has tripled within the last 25 years and the fraction of goods moved onto the road has substantially increased (up to two-thirds). The areas consist of small towns and villages, with a mix of industrial, small businesses, tourist, and agricultural activities. The primary noise sources are highway and railway traffic. In addition, densely trafficked main roads are of importance. These roads link the villages and towns and act as access roads to the highway.

Over the years, the sampling strategies have been refined. In the early studies all the people of the representative villages of a certain age range (25 - 65 years or 25 - 75 years) were approached by interviewers. In the later studies a basic phone survey (15 - 20 minutes) was conducted, based on a stratified, random sampling strategy. The address base was typically stratified using GIS (geographical information system) data, based on fixed distances, to the major traffic sources (railway, highway, main road), leaving a common 'background area' outside major traffic activities and an area with exposure to more than one traffic source ('mixed traffic area'). From these five areas, households were randomly selected and replaced in case of non-participation. Entry selection criteria were age range, sufficient hearing, language proficiency, and residency of at least one year at the current address. The participation was higher in the earlier survey (around 60%) and lower in the most recent survey (around 40%).

Noise exposure assessment

The earlier studies (Noise Village study, TRANSIT study) were based on the assessment of noise exposure on a short-term measurement network, with a central long-term recording unit. Subsequently, the individual noise exposure assignment was done in 5 dBA classes, based on these measurements, and local correction by noise expert judgments for each home. [4] No distinction was made between the contributing sources. In the Noise Village study there was a main road and a highway with a toll station. In the TRANSIT study in two of the five communities rail exposure was also of equal importance.

In the lower Inn valley EHIA-studies for the 'Brenner Eisenbahn Gesellschaft' (BEG studies: UIT-1, UIT-2) noise exposure (dBA, L dn ) was assessed by modeling (utilizing 'Soundplan TM ' software) and the calibration by measurements from 31 sites, according to Austrian guidelines (OAL Nr 28+30, ONORM S 5011). Based on both data sources approximate day-night levels (L dn ) were calculated for each respondent and also noise source, to facilitate comparison with the typical dose-response data. Exposure and survey data were then linked via GIS.

In the latest study (ALPNAP study), railway noise emission was extracted from a typical day of noise immission measurements at close distance to the source. For highway traffic the yearly average load (light and heavy vehicles) was combined with an average diurnal traffic pattern. For main roads, the available traffic frequency data were supplemented with additional traffic counting. Noise emission from road traffic was calculated with the help of the Harmonoise source model. [14] In addition, micro-simulations of the traffic flow were conducted with Paramics (Quadstone®, http://www.paramics-online.com ) to obtain optimal individual vehicle characteristics (speed and acceleration). Within the ALPNAP study for the first time two noise calculation procedures were implemented. 'Bass3', the propagation model developed by INTEC uses a three-dimensional object precise beam tracer, gradually becoming a stochastic ray tracer at a larger distance from the source, to determine the possible propagation paths. The sound propagation phenomena are included in an ISO9613-2 comparable manner. The model includes up to four reflections and two diffractions sideways. [15],[16] 'Mithra-Sig' is the implementation of the French NMPB-Routes-96 procedure by the Centre Scientifique et Technique du Bâtiment, Lyon (CSTB), of the current interim engineering methods recommended by the Environmental Noise Directive (END). It uses a 2.5 dimensional tracing for a visibility check. An extensive noise monitoring campaign was available to check the validity of these simulations. At 38 locations, the sound levels were recorded for over one week during winter (October to January) and during summer (June to August). In addition, the predicted sound pressure levels resulting from parabolic equation (PE)-modeling have been evaluated against these long-term measurements. [17] Indicators of day, evening, night exposure, and L den were calculated for each source, as also the total exposure at several points on the building facade of the survey participants. In the present analyses, L den at the façade most exposed was utilized.

Air pollution exposure assessment

In the BEG-studies, exposure by air pollution was assessed by a Swiss expert group (OEKOSCIENCE AG. Quellenstrasse 31, CH-8005 Zürich) who had long-term experience in monitoring and calibrating air pollution exposure in the alpine areas, with special consideration of meteorological conditions. An adapted Gaussian propagation modeling procedure was used. In the ALPNAP study, the annual means for NOx, NO 2 , and PM 10 were calculated for an area 27 km (W-E) × 23 km (N-S), east of Innsbruck). For these, air quality assessment of about 300 flow fields were calculated with the meteorological model GRAMM (Graz Mesoscale Model), [18],[19] for each domain. The model system used special algorithms to account for low wind or calm conditions. [19],[20] Traffic emissions were modeled using the network emission model NEMO. [21],[22] For each flow field a dispersion simulation was calculated with the Lagrangian particle model GRAL for horizontal resolutions of 10 × 10 m 2 and a vertical resolution of 2 m. [23],[24],[25] The model system used special algorithms to account for low wind or calm conditions. [19],[20] Each run was weighted due to its meteorological classification and frequency. Thereafter, annual, summer and winter means were calculated by post processing and weighting the numerous dispersion calculations. Within the ALPNAP study the simulation results were compared with seven air quality stations located in the Inn Valley. The background values within this study were height corrected according to Seinfeld & Pandis. [26] Calculated NO 2 and PM 10 values for each of the participant's home were assigned by GIS.

Questionnaire information

The questionnaire covered the sociodemographic data, housing, satisfaction with the environment, general noise annoyance, attitudes toward transportation, interference in activities, coping with noise, occupational exposures, lifestyle, reported sensitivities, health status, prevalent diseases, and intake of medications. The telephone interview took about 15 - 20 minutes to complete. Education was measured in five grades (basic, skilled, labor, vocational school, A-level, University degree). The last two grades were combined in the category 'higher education'. Noise sensitivity was queired with a five-point Likert-type question. 'High sensitivity' was defined by the two upper points on the scale (4 and 5). Health status was judged on a standard five-grade scale (1 to 5). The three poorest grades were combined as 'less than good' in the analysis. Active and emotional coping was assessed by a sum score, based on 13 items. [27] The area characteristics (urban, suburban, and rural) were defined by residential pattern and community size.

Statistical analysis

The statistical analyses were carried out with 'R' version 2.10. [28] Exposure-effect curves were calculated with extended logistics or ordinary least square regression methods using restricted cubic spline functions to accommodate for non-linear components in the fit, if appropriate. [29] In the results section the P values were reported for both the linear ('lin') and non-linear ('nlin') estimates. The non-parametric regression estimate and its 95% confidence intervals (CI) were based on smoothing the binary or continuous responses - in the case of binary response taking the logit transformation of the smoothed estimates - using the contributed R packages 'Design' and 'Hmisc'. [30] The criteria for the statistical consideration of interactions were relaxed, as departure from additivity could be of relevance in a public health context, when involved exposures and outcomes were prevalent. [31] It had also been demonstrated that selected studies could profit in terms of power by raising the Type I error rate from 5 to 20%, to detect interactions that would otherwise remain uncovered. [32],[33] This error rate was applied to report the 'relevant effect modification'. [Table 1] shows the major characteristics of the different studies.{Table 1}

 Results



Exposure-response relationship without consideration of effect modification

Statistically significant straight noise-effect relationships with basic adjustment of relevant confounders (no interactions) were only observed in selected analyses, with cardiovascular endpoints. In most analyses the noise-effect relation was only statistically significant in subgroups or in those with a predefined combination of susceptibility factors (mainly gender, age, family history of disease, and behavioral risk factors). To illustrate this point, first, only the exposure response relations of all studies that resulted from regression models with adjustment for standard factors without IA-terms were described. Note: The graphs show predicted probabilities based on modeled - not observed data.

In both the Noise Village study and the TRANSIT study no relevant relationship (main effect = ME) between noise and systolic blood pressure (SBP) could be observed. The UIT-1 study showed a slight linear relationship of hypertension with sound level, mainly in the older group [Figure 1]. The UIT-2 study exhibited a relationship of SBP with noise only in men at age 60. In the ALPNAP study, in both the hypertension and angina models, without interaction terms, a slight curve leveling off is visible around 60 dBA, L den [Figure 2]. Only in the UIT-1 study (basic hypertension model) the sound level increase between 50 and 60 dBA, L den , was significant (OR = 1.38, CI = 1.03 - 1.86). Furthermore, distance to the main road was a significant factor (P = 0.007). The companion models considering interactions are described a little later in the text, under the specific moderation heading. Interactions (IA) that were not significant in classical statistical terms were labeled as relevant effect modifications. In some studies we also described the relationship of distance to a relevant source. Note - the meaning of the air pollution models did not change when interaction terms were included.{Figure 1}{Figure 2}

Exposure-effect relationship with effect modification

a) Noise annoyance

It has been argued that subjective reports of actual perceived exposure are needed, in addition to objective indicators of noise, to better assess the potential adverse effects. [12],[34],[35] Due to the established noise-stress-CVD hypothesis of action, it would also seem reasonable to find noise-CVD associations, particularly among those who showed a particular disturbance or interference by noise, either during the day (impairment of concentration or performance) or night time (impairment of sleep). Only a few studies have tested these hypotheses. [34],[36] Overall, our data did not reveal any significant support to the simple hypothesis that higher noise annoyance is associated with higher cardiovascular disease outcome. To the contrary, from our early work in the Noise Village study we consistently observed the opposite in our SBP or hypertension relationships with traffic noise. Reporting higher annoyance (very much versus not at all) was significantly linked with lower SBP (- 5.83, CI = - 8.99 to - 2.68 mmHg), adjusting for age, sex, body mass index (BMI), education, cholesterol, family history, and window behavior in the TRANSIT study. [3] Likewise in the Noise Village study, the prevalence of hypertension was higher [Figure 3] in those reporting less interference by noise in their daily life (IA noise*interference P = 0.06).{Figure 3}

We explained this finding - which was unexpected at first glance - with the much higher adaptive efforts that higher annoyed subjects invested to reduce noise exposure compared with less annoyed subjects. [5] This supports a protective effect of certain active behavioral coping strategies - induced by higher annoyance. In the later studies, however, these associations of both coping activities and annoyance with blood pressure were weaker or no longer statistically significant. It remains to be speculated whether the health gain of active coping fades away over time when the troubling noise exposure situation persists. Alternatively, it could be that annoyance reporting habits changed over time or coping became more common. Thus, the power to detect health gains of protective behavior diminished over time.

b) Bedroom location

In the Tyrol studies we did not consistently observe the improved exposure effect relationships by either introducing bedroom location or an indicator of sleep disturbance as independent factors in the regression models. However, some models did improve. For example, participants with bedrooms facing toward a quiet yard [Figure 4] did show a clear trend toward a reduction in hypertension diagnoses in the ALPNAP-study (OR = 0.78, CI = 0.59 - 1.05). In the UIT-2 study a relevant interaction (IA) with bedroom location (IA: P = 0.18, ME bedroom : OR = 2.01(1.09, 3.78)) was observed when the distance to the main road was considered as an additional source parameter (ME distance: P = 0.02) in a non-significant rail noise model [Figure 5]. The interaction of bedroom location in the highway model was similar, but statistically not relevant (IA: P = 0.31) and also the single main effect (ME) of the bedroom location was less precise (ME bedroom: OR = 1.77, CI = 0.72, 4.39, ME distance : P = 0.08). In addition, the presence of night disturbance by rail did exhibit a further main effect in both the rail (OR = 2.24, CI = 1.21 - 4.17) and highway models (OR = 1.98, CI = 1.08 - 3.62).{Figure 4}{Figure 5}

c) Length of exposure

Duration of living at the current home may be another candidate variable representing a more homogeneous group with longer latency times for potential health effects. Bluhm et al., observed a stronger association between noise and hypertension in subjects with a longer period of residence (> 10 years). [37] We found no significant effect of longer duration on the overall noise-disease association in the ALPNAP-study, in the regression model. The duration of living was, however, strongly associated with older age (IA: P = 0.12) and house type (≥ 20 years in single homes 52%, versus 21% in apartment blocks). Thus, it is difficult to disentangle, especially when a large proportion of the sample has such a record of longer living (66%) or single housing (56%). However, when an extreme comparison was made with a strong family history of hypertension in the model adjustments, duration of living for ≥30 years at the present address was significantly associated with hypertension (OR = 1.68, CI = 1.07 - 2.66)) against < 8 years of living at that address. The comparison of living for ≥ 30 years versus < 30 years in the UIT-1 study revealed quite clear results supporting the theory that the length of exposure was an important variable [Figure 6]. In these analyses, the distance to the road was considered as exposure and heart problems as the outcome. When distance was replaced by the overall sound level, duration of living at the current home was significant (P = 0.04). However, no sign of effect modification by noise level was evident.{Figure 6}

d) Age

Hypertension: The re-analyses of the Tyrol health studies revealed substantial evidence for effect modification by age and gender on the relationship between noise and indicators of hypertension. Previously, in the small noise village study, we observed supporting evidence for a noise effect only in those at a higher age compared to participants at a lower age, with both dichotomous and continuous blood pressure outcomes. The interaction with noise level was statistically significant (IA: P = 0.01 lin, P = 0.02 nonlin). In the TRANSIT study, the age-noise level interaction on treated hypertension was only significant in men (IA treat: P = 0.03) [Figure 7]. In the UIT-1 study (IA: P = 0.02 lin, P = 0.03 nonlin) and the ALPNAP-study (IA diagnosis: P = 0.003 lin, P = 0.005 nonlin, IA treat: P = 0.013 lin, P = 0.006 nonlin) also, the overall effect modification by age was highly significant, with respect to any measurement of hypertension.{Figure 7}

Heart disease: In the TRANSIT-study we found a non-significant, but relevant, age-noise level interaction on the prevalence of angina pectoris, indicating that the relationship with noise only showed up in elderly people at higher noise levels [Figure 8]. This view was also supported in the results of the UIT-1 study when distance to the road was considered as exposure (IA: P = 0.26 lin). The analyses from the ALPNAP-study did not support these earlier findings. Rather, it was found that age was less important when other risk factors (e.g., hypertension) that are accompanied with older age were considered.{Figure 8}

e) Gender

Blood pressure and hypertension: In the Noise Village study the effect of the interaction of noise with age on systolic blood pressure was more pronounced in men [Figure 9]. In the TRANSIT study this kind of effect modification could be replicated with respect to the prevalence of hypertension (P = 0.22 lin). Also with a separate gender sub-regression the pattern was confirmed (IA age: men: P = 0.06, women: n.s). Likewise, a separate age sub-regression (three categories) on continuous blood pressure mimicked this pattern without reaching significance (IA sex-age: P = 0.33). Also in the UIT-2 study, interaction patterns due to gender (IA sex : P = 0.18 lin, 0.398 nonlin) occurred when adjustment for known hypertension was included. In the ALPNAP-study the effect of modification due to age was stronger - the noise-sex interaction, however, was of minor importance. In summary, we observed a stronger, although not always a significant effect of the noise level in men compared to women. In addition effect modification due to age was present and often enhanced the overall effect.{Figure 9}

Heart disease: Neither in the TRANSIT nor in the ALPNAP study did we find an indication of relevant effect modification due to gender on the relationship between the noise level and heart disease (IA: P = 0.6). In both the studies, the prevalence of angina pectoris was slightly higher among men across noise levels.

f) Education

Education was associated with both dichotomous and continuous blood pressure outcomes in the Noise Village study. The interaction of education with the noise level was evident in both sexes - but only relevant in the older age group (> 45 years) (IA: P = 0.20). The power (N = 174) was limited to test interactions. In the TRANSIT and the UIT-1 studies, education was not significant overall, due to social differences between the studied communities. In the UIT-2 study, a significant effect of education on systolic blood pressure was found (lower SBP in subjects with higher education compared to lower education: mean adjusted difference: - 3.90, CI = - 7.79 to - 0.01 mmHg) - with no relevant signs of interaction with noise level. In the ALPNAP study, this trend was reversed - but there hypertension diagnosis or treatment was the outcome. This might be due to differences in the detection and treatment of hypertension in general practice.

g) Family history

The TRANSIT study revealed some interaction of family history with noise level on SBP (IA: P = 0.11) in men, also in the presence of an additional effect modification by age (IA: P = 0.06 lin P = 0.14 nonlin). A similar non-significant result was obtained in the UIT 2 study with respect to systolic blood pressure. In fact the effect modification due to family history was caused by the interaction of sex with noise level. In the ALPNAP study we could test for possible interactions of family history with noise level with a more detailed question (no parent = 0, one parent = 1, two parents = 2). Two noise propagation models were tested. With the MITHRA propagation model the interaction with noise level was relevantly moderated by the degree of family history (IA: P = 0.11 lin) with an additional non-linear component [Figure 10]. Sex did not modify the associations any further, but age did, to a highly significant extent (IA: P = 0.013 lin, P = 0.006 nlin). Similar results were obtained when the International Organization for Standardization (ISO)-implementation by INTEC was used for noise propagation. The effect modification due to family history (IA: P = 0.16 lin) was obvious, but was not accompanied by a relevant effect modification, due to age. Further strong indications of interaction with family history were found with respect to other CV-outcomes when keeping age and sex constant or when including further risk factors.{Figure 10}

h) Hypertension

In the TRANSIT study [Figure 11] we observed a highly significant impact of known hypertension on the prevalence of angina pectoris (P ≤ 0.0001). However, no relevant interaction with the noise level was evident. Similar results were obtained in the ALPNAP study with respect to angina pectoris. [7] Subjects with pre-existing hypertension did exhibit a steeper increase in prevalence between noise levels of 50 and 60 dBA (OR = 2.23, CI = 1.10 - 4.52), but no statistically relevant effect modification of hypertension on the relationship between noise and angina pectoris was observed.{Figure 11}

i) Depression

In the TRANSIT study a borderline significant association between the prevalence of depression and the prevalence of angina pectoris was found. No interaction with the noise level was present. Although there was no association with blood pressure or hypertension in the ALPNAP study, we found a significant difference between people who suffered from depression and the probability of an angina pectoris diagnosis (OR = 2.06, CI = 1.08 - 3.94). There was, likewise, no relevant interaction with the noise level.

j) Air pollution

Hypertension: Support comes neither from the BEG studies nor from the ALPNAP study (not shown) for a significant positive effect of NO 2 or PM 10 on blood pressure or hypertension. Rather, opposite trends were observed. In addition no relevant signs of interaction could be found.

Angina pectoris: As in the case of hypertension, air pollution also did not affect the noise angina pectoris relation in the ALPNAP study. The observed inverse association is fully determined by the noise level.

k) Health status

A prospective study reported a stronger relation between annoyance and ischemic heart disease in middle-aged men with no prior disease at entry point. [38] A similar effect modification (P = 0.16) was observed in the presence of a strong noise*age interaction (P = 0.02) in the UIT-1 study, with respect to hypertension [Figure 12]. Only in those subjects with a good or very good health status a significant exposure effect relation was observed with regard to hypertension diagnosis or treatment. Also with respect to the three health status categories in the ALPNAP study, only for those with very good health status, a relation with noise level was found in men (IA: P = 0.18), but not in women. The ALPNAP study could not confirm such an effect modification of health status on the association between noise level and angina pectoris - although health status was a relevant predictor of disease when persons with excellent versus poor health status were compared (OR = 0.50, CI = 0.24 - 1.01).{Figure 12}

l) Combination of risk factors

Hypertension: Using the final model (adjusted for the other factors) of the ALPNAP study, simulations were carried out to demonstrate the relevant effect modification of the most important risk factors (age, family history, health status) on the relationship between the noise level and hypertension, when the factors were varied, in terms of extreme group comparisons [Figure 13].{Figure 13}

Heart disease: Likewise, we calculated the effect of two significant risk factors in the ALPNAP study, namely, hypertension and depression, on the probability of angina pectoris due to noise exposure (highway), for subjects of age 40 and 60 years, respectively [Figure 14]. A strong, effect modifying impact of the prevalence of these two diseases on the association between the noise level and the probability of developing angina pectoris was evident. However, the wide confidence intervals indicated the limitation when combinations with small subgroups were investigated.{Figure 14}

m) Noise sensitivity

Hypertension: With respect to hypertension, in none of the studies carried out in Tyrol were positive relations with noise sensitivity observed. To the contrary, noise sensitivity was consistently, non-significantly or even inversely associated with blood pressure readings or self-reported hypertension or treatment. On the other hand weather sensitivity (a general indicator of vegetative reactions) was a stronger predictor (P = 0.01) in the UIT-1 study - but here also no relevant interaction with noise level was evident [Figure 15]. The finding of higher weather sensitivity related to hypertension could not be fully replicated with systolic blood pressure as an outcome in the UIT-2 study. Instead, an underlying relevant sex-noise interaction (P = 0.17) was evident, only showing a noise effect in men. Noise sensitivity was again not a relevant parameter. However, vibration sensitivity also exhibited an inverse relation with blood pressure - but there was no effect modification of vibration sensitivity on the relationship between the noise level and blood pressure in the UIT-2 study. The smaller sample size in this study (N = 514) was the reason why only borderline significance was achieved (weather: P = 0.09, vibration: P = 0.05).{Figure 15}

Heart disease: Different results were obtained for heart disease. In the TRANSIT study, angina pectoris showed a non-significant association with noise sensitivity (P = 0.11), but there was neither any interaction with sex (IA: P = 0.98) nor with the noise level (IA: P = 0.72). In the UIT-1 study noise sensitivity was not a significant predictor either. Instead weather sensitivity exhibited a strong, effect modifying impact (IA: P = 0.11) on the relationship between the distance to the highway and the prevalence of angina pectoris [Figure 16]. In the ALPNAP study a different pattern was found [Figure 17]. A strong interaction of sensitivity with sex (IA: P = 0.01) was found on the non-linear relationship between the sound level (highway) and the prevalence of angina pectoris (IA: P = 0.16). The sex-sensitivity-interaction showed a deviating pattern: although sensitive males consistently showed the highest disease rates with varying noise levels, sensitive females exhibited the lowest rates of angina pectoris. Note - the confidence intervals are wide.{Figure 16}{Figure 17}

 Discussion



Exposure modifiers

The results from literature and the results of our studies over time suggest that there are important modifiers that may partly be responsible for the large variations found in the noise health effects research. Bluhm et al. suggest exposure misclassification as the main culprit. [37] Specifically, their findings of stronger associations in persons with a longer length of exposure (years of residence) at the same address, not having triple-glazed windows, bedroom windows directly facing a road or living in single houses, do support this suggestion of potential over- and underestimation of true exposure. Caution is warranted, as effect modification due to the length of exposure may actually be caused by older age, which is typically confounded with it.

On the other hand a longer duration of time spent living at the same address may also indicate a certain time of exposure required to exert an effect. [37] Therefore, studies with an insufficient proportion of people living longer at the same address (> 10 years) may lack the power to detect noise effects. In the Swedish study this proportion is high (44.5%). [37] In the ALPNAP study, 50% had lived for at least 16 years and 25% at least 30 years at the same address. From meta-analysis of annoyance studies we know that confounding with age is a serious problem. [38] In the Swedish study the age range went up to 80 years, [37] which is an unusually high age range, with inclusion of a large proportion of elderly people very likely to have lived longer at the same address. Another Swedish study with a sample age range up to 75 years also found a stronger association with > 10 years latency - but only for men - while the previous study observed a stronger association in women. [37]

We have seen protective effects (closing windows during night) as an additional modifier of exposure over and above the fact of having tightly fitted windows. [5] Tightly fitted windows or closing windows during daytime alone did not show up as significant variables.

Selander et al. found an elevated association between road traffic noise and myocardial infarction in participants reporting noise annoyance, mostly in their bedrooms. [36] These findings can be interpreted in different causal pathway directions. First, in general, bedroom exposure is a better exposure indicator by reducing exposure misclassification, as most participants (nightshift workers as an exception) are actually in bed, while daytime exposure can vary substantially due to activity pattern and work exposure. Second, bedroom exposure is a causally relevant exposure, as sleep is affected and impaired sleep is a known risk factor for myocardial infarction in men and women. [39],[40],[41],[42]

Our studies support bedroom location or night disturbance as a potential moderator especially when additional noise sources contribute to the overall noise exposure. The effect seems, however, to depend also on the kind of source combination (rail - highway - main road). Therefore, bedroom location should be considered in the analysis design - but high variation is possible due to the actual feature of the specific source combinations.

Effect modification: Socio-demographic factors

a) Gender and age

Several studies observed differences in the effect of noise on cardiovascular outcomes by gender. [37],[43],[44],[45],[46],[47],[48],[49] Unfortunately, the found associations are not uniform - thus casting a doubt on their reliability and validity. Similar to what has been argued in air pollution studies - that effects found only in women may be related to their longer duration of exposure at daytime - thus asserting that this issue is related to exposure assessment rather than implying a different vulnerability. However, there is evidence of a gender difference of psychophysiological reactions toward stress. Generally, males are more susceptible to cardiovascular disease [50] and women show greater resistance to stress between puberty and menopause. [51],[52]

In accordance with these findings the Tyrol studies did not provide support for a stronger effect in women. Instead, more often men did exhibit stronger effects in interactions with noise exposure and older age. When no effect modification by gender was observed, disparities in health care could be at work, like in hypertension treatment or angina pectoris diagnosis. [53],[54],[55],[56],[57] The extra-large studies of de Kluizenaar et al. and Bodin et al. used their power to test whether certain age ranges did exhibit stronger associations between noise and cardiovascular outcomes. [58],[59] The findings of these studies suggested the middle age ranges (40 - 60 years) to be associated with hypertension, but not other ages. As the findings reported so far concerned middle-aged people, explanations were targeted to explain this finding. In view of the results from the Tyrol studies, where the elderly were consistently more affected, other explanations were necessary and equally plausible. As noise was viewed as a subtle, but chronic stressor, longer latency periods might be necessary to observe the effects. In a recent, large, semi-ecological medication study in the same study area, we reported very significant findings for an age group above 70 years. [60]

The observed moderating effects of age or gender must be reviewed with caution, especially when only category-specific effects are reported and no exposure-effect relation is presented. Contrary to the argument of Bodin et al., [59] it can be stated that cohort analyses have shown that some classical cardiovascular risk factors lose their importance to predict cardiovascular mortality due to the survivor effect, and age gains importance. [61] Hence, the longer stress-related risks can exhibit their subtle effects, they can gain importance with age. Eventually, the support for a positive association between noise annoyance and cardiovascular health is weak. At least in the Tyrol studies, more often, the opposite effect was observed.

b) Education

The results show that when measured blood pressure was considered, lower education was consistently associated with higher blood pressure and higher prevalence of hypertension, based on the standard cut-off points. When reported hypertension was used, persons with a higher education exhibited a higher prevalence. This suggested a differential effect of health care on education. However, no significant effect modification with noise level was observed in any of the studies.

Effect modification: Vulnerability factors

a) Family history

Family history of hypertension is an established major risk factor for the development of hypertension. [62],[63],[64] In all studies (not available in UIT-1) family history was a significant contributor to either continuous or dichotomized blood pressure outcomes or treatment. Significant or a public health-relevant effect modification was observed with age and also with noise level. This supported the idea of higher vulnerability of people with a family history to noise exposure with a certain latency time. As more than one-third of the adult population in the ALPNAP study (41%) showed some degree of family history (one parent) effect modification should be evaluated in all noise - hypertension studies.

b) Hypertension

High blood pressure is a proven risk factor for cardiovascular diseases. [65] Selander et al. found a stronger association between road traffic noise and myocardial infarction in those with hypertension. [36] Earlier or recent hypertension was also a significant contributor to angina pectoris in the ALPNAP and the TERW-89 studies. The moderation with noise level did not become significant.

c) Depression

Depression is a known risk marker for cardiovascular diseases. [65] Most studies have found depression to be significantly associated with mortality and / or cardiac morbidity - although the mechanisms underlying this relationship remain unclear. [66],[67] As dysregulation of the autonomic nervous system is a plausible pathway to disease, chronic exposure to noise is a possible candidate for effect modification. [68] Both in the TRANSIT and the ALPNAP studies depressive symptoms or depression diagnosis were significant contributors in an angina pectoris regression model. Although some interaction between the noise level and the state of depression was visible in the figures, the power was too low to gain significance. However, the presence of both depression and hypertension showed a higher prevalence of angina at higher noise levels. We are not aware of other studies having evaluated depression as a possible moderator of the noise angina relationship.

d) Health status

Health status is a general and reliable predictor of future morbidity and mortality. [69],[70],[71],[72] In all the studies (not available in Noise village and TERW-89) health status has made a significant contribution to the cardiovascular outcomes studied. Consistently, persons with a poor health status have shown higher starting levels of morbidity, and typically also, stronger slopes in the exposure response analysis. However, due to the generally lower disease levels, sometimes only people with an excellent or good health status exhibit a significant increase of either hypertension or angina pectoris with increasing noise levels in a dose-response fashion. Therefore, effect modification has not always been significant at classical error rates (P < 0.05), but still relevant in terms of potential public health significance (P < 0.20). We are only aware of one study having applied a similar approach, by using the disease status as a possible moderator of the noise exposure disease relationship. [38]

e) Noise sensitivity

Noise sensitivity is known to be associated with higher symptom rates and medication consumption and is also a predictor of noise annoyance. [5],[73] In recent times, work based on data from the Finnish Twin Cohort study reported an association of noise sensitivity with hypertension, after adjustment for noise exposure and other factors, in a multivariate model. [74] In a further study, a relation between self-reported noise exposure and cardiovascular mortality was observed in noise sensitive women - but not among men. [75] On the other hand, we consistently observed a negative relationship between noise sensitivity and hypertension as a health endpoint. This was in total contrast to the results of the Finnish studies (overview in Heinonen 2009 [76] ) showing several associations of noise sensitivity with hypertension and heart disease (morbidity and mortality) in noise-exposed female subjects. Overall, there was a non-significant trend in noise-sensitive subjects to show a higher prevalence of angina pectoris at higher noise exposure - but due to a significant interaction (sex*sensitivity, P = 0.01) this was not true in women. Thus, there was no good evidence for a relation of noise sensitivity in women in this study. The power to detect weaker associations was low in the ALPNAP study. However, the pooled Caerphilly and Speedwell analyses (only men) also did not observe a significant association (OR = 0.9) with noise sensitivity in a larger sample. [77] Note, the Finnish studies differed methodologically, as both noise exposure and noise sensitivity was obtained subjectively. Such a procedure was vulnerable to the known subjective bias from stress research. [78],[79]

Measures of hypertension

In the noise literature, the clear diagnosis of hypertension (from medical sources or patient-remembered doctor diagnoses) is used in the analyses. Women are expected to show a lower prevalence of hypertension till the end of the fifth decade. [80] As medication use or type has not been confirmed in our studies, misclassification may be introduced by other unknown medications that may lower blood pressure. Furthermore, true awareness and control rates cannot be determined with the kind of data available. The literature reports awareness rates to be around 70%. Treatment and control rates are found to be around 60 and 30%, respectively, with lower rates in the elderly. [80],[81] The experience with other surrogate measures of hypertension in our studies show the following characteristics:



Blood pressure readings are less often significantly related to noise levelsDichotomizing blood pressure readings at higher cut-off levels (160 / 95 mmHg) are more likely significantly associated with noise exposureTreated hypertension is not a better indicator than doctor diagnosis or known hypertensionWhen using treated hypertension we have not observed a gender difference in prevalence. This gender difference is consistently present when using remembered diagnosis or personal readings of blood pressure - indicating a lower prevalence of treatment among men than women. These findings are confirmed by large population surveys - but it seems that the male population is catching up. [81]

Time effects and latency to the effect

As these are series of cross-sectional studies over time, it is difficult to comment on time factors. However, there are some findings that contribute to the current scarce knowledge:

In the studies where we had two time frames in the retrospective question available (e.g., 'hypertension diagnosis ever' and 'hypertension diagnosis during past 12 months') the precise time framing 'past 12 months' did exhibit a stronger relation with noise than the more loose time framing 'ever'. This finding may be explained by the concurrent measurement of exposure and outcome, and thus, reflect a higher precision in both. Alternatively, it could also give hints for time windows, where a certain proportion of the study population may exhibit noise-related effects.Consistently, we found that persons at a higher age (> 60 years) showed a firmer relation with noise than those at a lower age (~ 40 years). This relation was typically enhanced in the presence of an additional risk factor for the outcome under investigation (especially family history of hypertension). These findings suggest longer latency times and the need for other risk factors to be present in order to develop noise-related effects. The findings from our semi-ecological study, where significant relations with noise (antihypertensive prescriptions) were only found at an older age (> 70 years), indirectly support longer latency times in general. [60] Duration of living may not be a good approximation of the length of exposure, as it is strongly associated with age, housing factors, and education in our studies. From the social science literature it is long known and recently confirmed that people moving around less are better off in the light of various health outcomes and health-related behavior. [82],[83],[84],[85] At least in our studies we have observed no significant difference in subset analyses with 10, 20 or 30 years of living at the same address. Although, utilizing the length of residence as a continuous variable with adjustments of age, housing, education, and health status does show a small increase in the odds for hypertension development when the contrast in duration is stretched (< 8 years versus > 30 years), it is not clear whether this indeed represents an independent finding. However, it also supports longer latency times, similar to the conclusions from the Speedwell and Caerphilly studies (>15 years).

Air pollution

Both noise and air pollution are often emitted by the same source, namely motorized traffic, and depending on the propagation conditions a wide range of correlations is reported. [86] Such conditions open the possibility of confounding and make it difficult to disentangle the associated effects statistically. [87] A large number of studies have shown stable associations of ambient air pollution with morbidity and mortality of cardiopulmonary disease. [88],[89],[90] A smaller number of studies have reported associations with blood pressure or hypertension. [91],[92] As noise exposure is also associated with coronary heart disease (CHD) and hypertension, [93] and only a few recent studies have actually considered both pollutants in the regression models, [7],[36],[58],[94] it remains an open question as to what contribution is made by which pollutant to which health outcome.

In both the UIT studies and the ALPNAP study high quality air and noise pollution propagation data were available for individual assignment. In none of the investigated health endpoints (angina, blood pressure / hypertension) a relevant or consistent relation with the studied range of air pollutants (NO 2 , PM 10 ) nor a relevant moderation could be established. The large population-based Oslo Health Study (N = 18,770) was also unable to find a relation between the indicators of air pollution exposure and blood pressure. [95] As we had two noise assignment options in the ALPNAP study, of which the ISO-assignments showed very low correlations with NO 2 (r = 0.12) and PM 10 (r = 0.09), confounding was highly unlikely to be of importance in this study. Although the MITHRA-assignments showed higher correlations (NO 2 : r = 0.48 and PM 10 : r = 0.39), the statistical importance of both the air pollutants and the noise variables did not change. In the BEG studies the highway noise to air correlations were stronger (NO 2 : r = 0.63 and PM 10 : r = 0.61).

Methodological issues

a) Interaction assessment and non-linearity

The investigation of moderation of the noise health relationship by public health relevant factors is a necessary requirement for the better understanding of the processes that determine the person-environment-health relationship. [12],[59],[96],[97] A single reporting of the average risk effects or associations from an entire population can often conceal the substantial variation that may occur in important subgroups (the elderly, women, and persons with positive family history of cardiovascular diseases, including high blood pressure). This deviation from the average risk can be even more pronounced when risk combinations are considered. If significant interactions are present, the meaning of the main effects becomes questionable. Unfortunately, most studies do not have the power to evaluate effect modification and interaction tests, in general, lack power. [32],[98] The relaxation of the significance criteria can sometimes help. [32],[33] The use of P values as the sole criterion is discouraged. [99] However, caution is needed as additional mediation or residual confounding may distort the results or make it difficult to interpret. [100] Therefore, only a biological plausible effect modification (based on prior knowledge) must be tested and a step-by-step procedure is advised - followed by detailed sensitivity analyses, to safeguard the conclusions. Eventually, there is a strong need to examine non-linear components in the exposure-response analyses. Substantial over- and underestimations may result without consideration.

b) Noise propagation modeling

Typically, engineering methods and the resulting noise maps are validated against long-term noise measurements in 'simple' open area propagation conditions and not in complex residential settings, where most people actually live. The availability of having two (in the case of highway noise, three) noise propagation methods in the ALPNAP study opened the unique opportunity to evaluate the modeling in the framework of actual noise - health relations. Thus, the effects of noise modeling techniques on the estimation of noise-associated health impacts could be directly assessed. Although sometimes only marginal differences were noted even with complex effect modification, in other cases (e.g., angina pectoris) with only one method a significant exposure-response relationship was established, but not with the second method. This leaves behind a substantial amount of uncertainty. Therefore, a move from mere exposure modeling to exposure effect modeling is required to minimize the bias in public health risk assessment of the effects of sound on humans.

 Conclusions



Because noise is not a strong risk factor per se, the specific context of the exposure, health predispositions, and the adaptability to this person-environment configuration determines whether effects occur. Specifically, the coping opportunities are of importance. If active coping (closing windows, bedroom on quiet site) is not feasible noise persists as a chronic stressor and with advancing age the effects may surface. As the effects of age and gender observed in noise effects research can only be prevented by reducing the intensity and the duration of exposure overall - residential areas should be considered as sensitive areas and the noise here must not exceed 55 dBA. This is in accordance with the results of the most recent studies. [36],[37],[49] Finally, from the reported studies we have not been able to find support for a relevant role of air pollution. Neither with hypertension nor with heart disease a statistical significant association did come in reach. Eventually, no signs for a relevant moderation of the noise health relation by air pollution could be observed in the Tyrol studies.

 Acknowledgments



The ALPNAP project received European Regional Development Funding through the INTERREG III Community Initiative. In the context of this study we received data and supporting information from various governmental, private, and public institutions. Special thanks to the GEO-information system TIRIS and the traffic administration of the Tyrolean Government and the Brenner Railway Company (BEG). The phone surveys were conducted in all studies by the CAT-Lab of IMAD, an experienced opinion research Institute in Innsbruck. We thank the study participants and greatly acknowledge the field work done in the interview studies by various students, doctoral students from the medical and psychology curricula, and supervising MDs and PhDs. The study was partly supported by the Department / Division of Social Medicine, Medical University Innsbruck, formerly The University of Innsbruck. Parts of the analyses were conducted for the ENNAH EC 7 th Framework research programme.

References

1Lercher P. Straßenverkehr und Gesundheit. Das Beispiel "Lärmdorf". Innsbruck: Veröffentlichungen der Universität Innsbruck Nr. 166; 1988. p. 27-48.
2Lercher P. Community exposure to highway noise, behavioural adaptation and blood pressure. Master′s Thesis. University of North Carolina at Chapel Hill, 1990.
3Lercher P. Auswirkungen des Straßenverkehrs auf Lebensqualität und Gesundheit. (Effects of road traffic on quality of life and health) Transitstudie - Sozialmedizinischer Teilbericht. Innsbruck: Amt der Tiroler Landesregierung; 1992.
4Lercher P, Schmitzberger R, Kofler W. Perceived traffic air pollution, associated behaviour and health in an alpine area. Sci Total Environ 1995;169:71-4.
5Lercher P, Kofler WW. Behavioural and health responses associated with road traffic noise along alpine through-traffic routes. Sci Total Environ 1996a;189/190:85-9.
6Lercher P, Widman U, Kofler W. Transportation noise and blood pressure: the importance of modifying factors: In: Cassereau D, editor. Proceedings of Internoise 2000, Vol. 4, Nice: Societe Francaise d′Acoustique; 2000. p. 2071-5.
7Heimann D, de Franceschi M, Emeis S, Lercher P, Seibert P. Air Pollution, Traffic Noise and Related Health Effects in the Alpine Space - A Guide for Authorities and Consulters. ALPNAP comprehensive report. Università degli Studi di Trento, Trento, Italy, Dipartimento di Ingegneria Civile e Ambientale; 2007. Available from: http://www.alpnap.org/alpnap.org_ge.html [cited in 2007].
8Cohen S, Evans GW, Stokols D, Krantz, DS. Behaviour, health and environmental stress. New York, London: Plenum Press; 1986.
9Lercher P. Non-auditory health effects research: A conceptual analysis of recent results and research needs. In: Kuwano S, editor. Proceedings of Internoise 94, Yokohama, 1994. p. 1053-8.
10Lercher P, Stansfeld SA, Thompson SJ. Non-auditory health effects of noise: review of the 1993-8 period. In: Carter NL, Job RF, editors. Noise as a public health problem. Sydney: Noise Effects ′98 Pty Ltd; 1998b. p. 213-20.
11Lercher P. Environmental noise: A contextual public health perspective. In: Luxon L, Prasher D, editors. Noise and its effects: London: Wiley; 2007. p. 345-77.
12Lercher P. Environmental noise and health: An integrated research perspective. Environ Intern 1996b;22:117-29.
13Lercher P. Context and coping as moderators of potential health effects in noise-exposed persons. In: Prasher D, Luxon L, editors. Advances in noise series. Vol 1: biological effects. London: Whurr; 1998a. p. 328-35.
14Jonasson HG. Acoustical Source Modelling of Road Vehicles, Acta Acustica united with Acustica 2007;93:174-83.
15de Greve B, de Muer T, Botteldooren D. Outdoor beam tracing over undulating terrain. In: Augusztinovicz F, Nagy AB, Hunyadi Z, editors. Proceedings of Forum Acusticum. Budapest: OPAKFI Tudomanyos Egyesület; 2005. on CD.
16de Greve B, van Renterghem T, Botteldooren D. Outdoor sound propagation in mountainous areas: comparison of reference and engineering models. In: Calco-Manzano A, Pérez-López A, Santiago JS, editors. Proceedings of the 19th International Congress on Acoustics. Madrid: Spanish Acoustical Society; 2007. on CD.
17Van Renterghem T, Botteldooren D, Lercher P. Comparison of measurements and predictions of sound propagation in a valley-slope configuration in an inhomogeneous atmosphere. J Acoust Soc Am 2007;121:2522-33.
18Almbauer R. A, Öttl D, Bacher M, Sturm PJ. Simulation of the air quality during a field study for the city of Graz. Atmos Environ 2000;34:4581-94.
19Öttl D, Goulart A, Degrazia G, Anfossi D. A new hypothesis on meandering atmospheric flows in low wind speed conditions. Atmos Environ 2005;39:1739-48.
20Öttl D, Almbauer R, Sturm PJ. A new method to estimate diffusion in low wind, stable conditions. J Appl Meteorol 2001;40:259-68.
21Rexeis M, Hausberger S. Calculation of Vehicle Emissions in Road Networks with the Model "NEMO". Conference Proceedings of the 14th Symposium on Transport and Air Pollution. ISBN 3-902465. Graz: 2005. p. 118-28.
22Rexeis M, Hausberger S. Trend of vehicle emission levels until 2020, prognosis based on current vehicle measurements and future emission legislation. Atmos Environ 2009;43:4689-98.
23Öttl D, Sturm P, Pretterhofer G, Bacher M, Rodler J, Almbauer R. Lagrangian dispersion modeling of vehicular emissions from a highway in complex terrain. J Air Waste Manag Assoc 2003a;53:1233-40.
24Öttl D, Almbauer RA, Sturm PJ, Pretterhofer G. Dispersion modelling of air pollution caused by road traffic using a Markov Chain - Monte Carlo model. Stoch Environ Res Risk Assess 2003b;17:58-75.
25Öttl D, Sturm P, Anfossi D, Trini Castelli S, Lercher P, Tinarelli G, et al. Lagrangian particle model simulation to assess air quality along the Brenner transit corridor through the Alps; Air Pollution Modelling and its Applications XVIII; Developments in Environmental Science 6, New York: Elsevier; 2007. p. 689-7.
26Seinfeld JH, Pandis SN. Atmospheric chemistry and physics: From air pollution to climate change. ISBN 0-471-17815-2. New York: Wiley-Interscience; 1997.
27Botteldooren D, Lercher P. Soft-computing base analyses of the relationship between annoyance and coping with noise and odor. J Acoust Soc Am 2004;115:2974-85.
28R Development Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing. Vienna, Austria; 2009. ISBN 3-900051-07-0. Available from: http://www.R-project.org. [cited in 2009].
29Harrell FE Jr. Regression modelling strategies. New York: Springer; 2001.
30Harrell FE Jr. R-Hmisc and R-Design library. Available from: http://www.cran.r-project.org/web/packages/Design/
31Greenland S, Rothman KJ. Chapter 18: Concepts of Interaction. In: Rothman KJ, Greenland S, editors. Modern Epidemiology. 2nd ed. New York: Lippincott-Raven; 1998. p. 329-42.
32Marshall SW. Power for tests of interaction: effect of raising the Type I error rate. Epidemiologic Perspectives and Innovations; 2007.
33Selvin S. Statistical Analysis of Epidemiologic Data New York: NY: Oxford University Press; 1996. p. 213-4.
34Babisch W, Gallacher J, Ising, H . Schallpegel oder subjektive Störung. Lärmexpositionsmaße in Wirkungsstudien am Beispiel einer Kohortenstudie. (In German). Bundesgesundheitsblatt 1995;4:137-45.
35Fields JM. Effect of personal and situational variables on noise annoyance in residential areas. J Acoust Soc Am 1993;93:2753-63.
36Selander J, Nilsson ME, Bluhm G, Rosenlund M, Lindqvist M, Nise G, et al. Long-term exposure to road traffic noise and myocardial infarction. Epidemiology 2009;20:272-9.
37Bluhm GL, Berglind N, Nordling E, Rosenlund M. Road traffic noise and hypertension. Occup Environ Med 2007;64:122-6.
38Babisch W, Ising H, Gallacher JE. Health status as a potential effect modifier of the relation between noise annoyance and incidence of ischaemic heart disease. Occup Environ Med 2003;60:739-45.
39Schwartz S, McDowell Anderson W, Cole SR, Cornoni-Huntley J, Hays JC, Blazer D. Insomnia and heart disease: a review of epidemiologic studies. J Psychosom Res 1999;47:313-33.
40Greenland P, Knoll MD, Stamler J, Neaton JD, Dyer AR, Garside DB, et al. Major risk factors as antecedents of fatal and nonfatal coronary heart disease events. JAMA 2003;290:891-7.
41Leineweber C, Kecklund G, Janszky I, Akerstedt T, Orth-Gomer K. Poor sleep increases the prospective risk for recurrent events in middle-aged women with coronary disease. The Stockholm female coronary risk study. J Psychosom Res 2003;54:121-7.
42Meisinger C, Heier M, Löwel H, Schneider A, Döring A. Sleep duration and sleep complaints and risk of myocardial infarction in middle-aged men and women from the general population: the MONICA/KORA Augsburg cohort study. Sleep 2007;30:1121-7.
43Herbold M, Hense HW, Keil U. Effects of road traffic noise on prevalence of hypertension in men: Results of the Lübeck blood pressure study. Soz Praeventiv Med 1989;34:19-23.
44Belojevic G, Saric-Tanaskovic M. Prevalence of arterial hypertension and myocardial infarction in relation to subjective ratings of traffic noise exposure. Noise Health 2002;4:33-7.
45Babisch W, Beule B, Schust M, Kersten N, Ising H. Traffic noise and risk of myocardial infarction. Epidemiology 2005;16:33-40.
46Willich SN, Wegscheider K, Stallmann M, Keil T. Noise burden and the risk of myocardial infarction. Eur Heart J 2006;27:276-82.
47Bjork J, Ardo J, Stroh E, Lovkvist H, Ostergren PO, Albin M. Road traffic noise in southern Sweden and its relation to annoyance, disturbance of daily activities and health. Scand J Work Environ Health 2006;32:392-401.
48Jarup L, Babisch W, Houthuijs D, Pershagen G, Katsouyanni K, Cadum E, et al. Hypertension and exposure to noise near airports - the HYENA study. Environ Health Perspect 2008;116:329-33.
49Barregard L, Bonde E, Öhrström E. Risk of hypertension from exposure to road traffic noise in a population-based sample. Occup Environ Med 2009;66:410-5.
50Stoney CM, Davis MC, Matthews KA. Sex differences in physiological responses to stress and in coronary heart disease: A causal link? Psychophysiology 1987;24:127-31.
51Kajantie E, Phillips DIW. The effects of sex and hormonal status on the physiological response to acute psychosocial stress. Psychoneuroendocrinology 2006;31:151-78.
52Kajantie E. Physiological stress response, estrogen, and the male-female mortality gap. Current Directions in Psychological Science 2008;17:348-52.
53Vaccarino V. Angina and cardiac care: are there gender differences, and if so, why? Circulation 2006;113:467-9.
54Crilly M, Bundred P, Hu X, Leckey L, Johnstone F. Gender differences in the clinical management of patients with angina pectoris: a cross-sectional survey in primary care. BMC Health Serv Res 2007;7:142.
55Gu Q, Burt VL, Paulose-Ram R, Dillon CF. Gender differences in hypertension treatment, drug utilization patterns, and blood pressure control among US adults with hypertension: data from the National Health and Nutrition Examination Survey 1999-2004. Am J Hypertens 2008;21:789-98.
56Bittner V. Angina pectoris. reversal of the gender gap. Circulation 2008;117:1505-7.
57Hemingway H, Langenberg C, Damant J, Frost C, Pyörälä K, Barrett-Connor E. Prevalence of angina in women versus men: a systematic review and meta-analysis of international variations across 31 countries. Circulation 2008;117:1526-36.
58de Kluizenaar Y, Gansevoort RT, Miedema HM, de Jong PE. Hypertension and road traffic noise exposure. J Occup Environ Med 2007;49:484-92.
59Bodin T, Albin M, Ardö J, Stroh E, Ostergren PO, Björk J. Road traffic noise and hypertension: results from a cross-sectional public health survey in southern Sweden. Environ Health 2009;8:38.
60Rüdisser J, Lercher P, A Heller. Traffic exposure and medication - a GIS based study on prescription of medicines in the Tyrolean Wipptal. Ital J Public Health 2008;5:261-7.
61Grundy SM, D′Agostino RB Sr, Mosca L, Burke GL, Wilson PW, Rader DJ, et al. Cardiovascular risk assessment based on US cohort studies: findings from a National Heart, Lung, and Blood institute workshop. Circulation 2001;104:491-6.
62Stamler R, Stamler J, Riedlinger WF, Algera G, Roberts RH. Family (parental) history and prevalence of hypertension. JAMA 1979;241:43-6.
63Burke V, Gracey MP, Beilin LJ, Milligan RA. Family history as a predictor of blood pressure in a longitudinal study of Australian children. J Hypertens 1998;16:269-76.
64Wang NY, Young JH, Meoni LA, Ford DE, Erlinger TP, Klag MJ. Blood pressure change and risk of hypertension associated with parental hypertension: The Johns Hopkins Precursors Study. Arch Intern Med 2008;168:643-8.
65Yusuf S, Reddy KS, Ôunpuu S, Anand S. Global burden of cardiovascular diseases part I: general considerations, the epidemiologic transition, risk factors, and impact of urbanization. Circulation 2001;104:2746-53.
66Suls J, Bunde J. Anger, anxiety, and depression as risk factors for cardiovascular disease: the problems and implications of overlapping affective dispositions. Psychol Bull 2005;131:260-300.
67Carney RM, Freedland KE, Veith RC. Depression, the autonomic nervous system, and coronary heart disease. Psychosom Med 2005;67:29-33.
68Carney RM, Freedland KE, Miller GE, Jaffe AS. Depression as a risk factor for cardiac mortality and morbidity: a review of potential mechanisms. J Psychosom Res 2002;53:897-902.
69Idler EL, Benyamini Y. Self-assessed health and mortality: a review of twenty seven community studies. J Health Soc Behav 1997;38:21-37.
70Lekander M, Elofsson S, Neve IM, Hansson LO, Unden AL. Self-rated health is related to levels of circulating cytokines. Psychosom Med 2004;66:559-63.
71Chen H, Cohen P, Kasen S. Cohort differences in self-rated health: evidence from a three-decade, community-based, longitudinal study of women. Am J Epidemiol 2007;166:439-46.
72Singh-Manoux A, Guéguen A, Martikainen P, Ferrie J, Marmot M, Shipley M. Self-rated health and mortality: Short- and long-term associations in the Whitehall II study. Psychosom Med 2007;69:138-43.
73Stansfeld SA. Noise, noise sensitivity and psychiatric disorder: epidemiological and psychological studies. Psychological medicine. Monograph Supplement 22. Cambridge: Cambridge University Press; 1992.
74Heinonen-Guzejev M, Vuorinen HS, Kaprio J, Heikkilä K, Mussalo-Rauhamaa H, Koskenvuo M. Transportation noise exposure, annoyance and noise sensitivity in relation to noise map information. J Sound Vibration 2004;234:191-206.
75Heinonen-Guzejev M, Vuorinen HS, Mussalo-Rauhamaa H, Heikkilä K, Koskenvuo M, Kaprio J. The association of noise sensitivity with coronary heart and cardiovascular mortality among Finnish adults. Sci Total Environ 2007;372:406-12.
76Heinonen-Guzejev M, Vuorinen HS, Mussalo-Rauhamaa H, Heikkilä K, Koskenvuo M, Kaprio J. Medical, psychological and genetic aspects of noise sensitivity. In: Proceedings of the 8th European Conference on Noise Control. Edinburgh: Institute of Acoustics on behalf of the EEA; 2009. on CD.
77Babisch W, Ising H, Gallacher JE, Sweetnam PM, Elwood PC. Traffic noise and cardio-vascular risk: The Caerphilly and Speedwell studies, third phase - 10 years follow-up. Arch Environ Health 1999;54:210-6.
78Kasl SV. Stress and health. Ann Rev Publ Health 1984;5:319-41.
79Lazarus RS. Coping theory and research: past, present, and future. Psychosom Med 1993;55:234-47.
80Hajjar I, Kotchen TA. Trends Hajjar I, Kotchen TA. Trends in prevalence, awareness, treatment, and control of hypertension in the United States, 1988-2000. JAMA 2003;290:199-206.
81Cutler JA, Sorlie PD, Wolz M, Thom T, Fields LE, Roccella EJ. Trends in hypertension prevalence, awareness, treatment, and control rates in United States Adults Between 1988-1994 and 1999-2004. Hypertension 2008;52:818-27.
82Metzner HL, Harburg E, Lamphiear DE. Residential mobility and urban-rural residence within life stages related to health risk and chronic disease in Tecumseh, Michigan. J Chronic Dis 1982;35:359-74.
83Larson A, Bell M, Young A. Clarifying the relationships between health and residential mobility. Social Science and Medicine 2004;59:2149-60.
84Norman P, Boyle P, Rees P. Selective migration, health and deprivation: a longitudinal analysis. Soc Sci Med 2005;60:2755-71.
85Jelleyman T, Spencer N. Residential mobility in childhood and health outcomes: a systematic review. J Epidemiol Community Health 2008;62:584-92.
86Allen RW, Davies H, Cohen MA, Mallach G, Kaufman JD, Adar SD. The spatial relationship between traffic-generated air pollution and noise in 2 US cities. Environ Res 2009;109:334-42.
87Schwela D, Kephalopoulos S, Prasher D. Confounding or aggravating factors in noise-induced health effects: air pollutants and other stressors. Noise Health 2005;7:41-50.
88Pope CA, Dockery DW. Health effects of fine particulate air pollution: lines that connect. J Air Waste Manag Assoc 2006;56:709-42.
89Krewski D, Rainham D. Ambient air pollution and population health: overview. J Toxicol Environ Health A 2007;70:275-83.
90Brook RD. Cardiovascular effects of air pollution. Clin Sci 2008;115:175-87.
91Ibald-Mulli A, Stieber J, Wichmann HE, Koenig W, Peters A. Effects of air pollution on blood pressure: A population-based approach. Am J Public Health 2001;91:571-7.
92Zanobetti A, Canner MJ, Stone PH, Schwartz J, Sher D, Eagan-Bengston E, et al. Ambient pollution and blood pressure in cardiac rehabilitation patients. Circulation 2004;110:2184-9.
93Babisch W. Road traffic noise and cardiovascular risk. Noise Health 2008;10:27-33.
94Beelen R, Hoek G, Houthuijs D, van den Brandt PA, Goldbohm RA, Fischer P, et al. The joint association of air pollution and noise from road traffic with cardiovascular mortality in a cohort study. Occup Environ Med 2009;66:243-50.
95Madsen C, Nafstad P. Associations between environmental exposure and blood pressure among participants in the Oslo Health Study (HUBRO). Eur J Epidemiol 2006;21:485-91.
96Evans GW, Lepore SJ. Moderating and mediating processes in environment-behavior research. In: Moore GT, Morons RW, editors. Advances in environment, behavior, and design. Vol 4. New York: Plenum Press; 1997. p. 255-85.
97Evans GW. Environmental stress and health. In: Baum A, Revenson T, Singer JE, editors. Handbook of health psychology. Mahwah, NJ: Erlbaum; 2001: pp365-85.
98Greenland S. Basic problems in interaction assessment. Environ Health Perspect 1993;101:59-66.
99Matthews JN, Altman DG. Education and debate Statistics Notes: Interaction 2: compare effect sizes not P values. BMJ 1996;313:808.
100Pearce N, Greenland S. Confounding and interaction. In: Ahrens W, Krickeberg K, Pigeot I, editors. Handbook of epidemiology. Heidelberg: Springer-Verlag; 2004. p. 375-401.