• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptNIH Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Environ Res. Author manuscript; available in PMC Apr 6, 2011.
Published in final edited form as:
PMCID: PMC3071639

The spatial relationship between traffic-generated air pollution and noise in 2 US cities[star]


Traffic-generated air pollution and noise have both been linked to cardiovascular morbidity. Since traffic is a shared source, there is potential for correlated exposures that may lead to confounding in epidemiologic studies. As part of the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air), 2-week NO and NO2 concentrations were measured at up to 105 locations, selected primarily to characterize gradients near major roads, in each of 9 US communities. We measured 5-min A-weighted equivalent continuous sound pressure levels (Leq) and ultrafine particle (UFP) counts at a subset of these NO/NO2 monitoring locations in Chicago, IL (N = 69 in December 2006; N = 36 in April 2007) and Riverside County, CA (N = 46 in April 2007). Leq and UFP were measured during non-“rush hour” periods (10:00–16:00) to maximize comparability between measurements. We evaluated roadway proximity exposure surrogates in relation to the measured levels, estimated noise–air pollution correlation coefficients, and evaluated the impact of regional-scale pollution gradients, wind direction, and roadway proximity on the correlations. Five-minute Leq measurements in December 2006 and April 2007 were highly correlated (r = 0.84), and measurements made at different times of day were similar (coefficients of variation: 0.5–13%), indicating that 5-min measurements are representative of long-term Leq. Binary and continuous roadway proximity metrics characterized Leq as well or better than NO or NO2. We found strong regional-scale gradients in NO and NO2, particularly in Chicago, but only weak regional-scale gradients in Leq and UFP. Leq was most consistently correlated with NO, but the correlations were moderate (0.20–0.60). After removing the influence of regional-scale gradients the correlations generally increased (Leq–NO: r = 0.49–0.62), and correlations downwind of major roads (Leq–NO: r = 0.53–0.74) were consistently higher than those upwind (0.35–0.65). There was not a consistent effect of roadway proximity on the correlations. In conclusion, roadway proximity variables are not unique exposure surrogates in studies of endpoints hypothesized to be related to both air pollution and noise. Moderate correlations between traffic-generated air pollution and noise suggest the possibility of confounding, which might be minimized by considering regional pollution gradients and/or prevailing wind direction(s) in epidemiologic studies.

Keywords: Air pollution, Noise, Traffic, Confounding, Cardiovascular

1. Introduction

Researchers have reported associations between chronic exposure to traffic and adverse cardiovascular health effects including hypertension, myocardial infarction, stroke, atherosclerosis, heart disease, and mortality. These associations have been attributed to traffic-generated air pollution (Finkelstein et al., 2004; Hoek et al., 2002; Hoffmann et al., 2006, 2007; Maheswaran and Elliott, 2003; Tonne et al., 2007) or road noise (Babisch, 2006; Babisch et al., 2005; Bluhm et al., 2007; de Kluizenaar et al., 2007; Selander et al., 2009; van Kempen et al., 2002). If air pollution and noise are both linked to cardiovascular effects, the fact that traffic is a major shared source suggests the potential for correlated exposures that may lead to confounding in epidemiologic studies (Schwela et al., 2005).

The potential for confounding is increased by the exposure assessment approaches that are commonly used in epidemiologic studies, in which it is not feasible to measure exposure for every participant. As an alternative to measurements, investigations of road noise and health generally use models to estimate noise exposures (Babisch et al., 2005; Beelen et al., 2008; Bluhm et al., 2007; Calixto et al., 2003; de Kluizenaar et al., 2007; van Kempen et al., 2002). Similarly, some studies of traffic-generated air pollution use dispersion and/or “land use regression” models to estimate concentrations (Ainslie et al., 2008; Jerrett et al., 2005a; Su et al., 2008). However, these air pollution models often require spatially dense monitoring or extensive data on emissions and meteorology prior to model development. As a result, simple roadway proximity-based metrics are commonly used as exposure surrogates, in part because they are easily implemented using readily available data and do not require any air pollution measurements (Adar and Kaufman, 2007; Finkelstein et al., 2004; Hoek et al., 2002; Hoffmann et al., 2007, 2006; Jerrett et al., 2005a, b; Maheswaran and Elliott, 2003; Tonne et al., 2007). These surrogate measures are based on correlations between roadway proximity and measured levels of traffic-generated air pollutants (Beckerman et al., 2008; Gilbert et al., 2007, 2003; Pleijel et al., 2004; Roorda-Knape et al., 1998; Zhu et al., 2002). However, interpretation of epidemiologic studies that use the proximity approach is complicated by the fact that noise levels are also related to roadway proximity (Hothersall and Chandler-Wilde, 1987). The ability of these roadway proximity metrics to predict measured levels of air pollution and noise has not been directly compared.

The published data on the relationship between noise and air pollution are also very limited and somewhat inconsistent. A study in Madrid evaluated the relationship between 1096 daily measurements of noise (measured at 6 locations) and NO2/NOx (measured at 24 locations) (Tobias et al., 2001). The authors reported noise-NO2 and noise-NOx correlation coefficients of 0.32 and 0.35, respectively. However, while informative for daily time-series studies, these temporal relationships are of limited value in interpreting epidemiologic studies of chronic exposures in which the spatial exposure contrast is of interest. More relevant to chronic effects studies are the findings of Klaeboe et al. (2000), who modeled 24-h Leq and 3-month average NO2 concentrations based on traffic volumes at approximately 1000 locations in Oslo and reported a modest relationship (r = 0.46). In a study of chronic noise exposure and hypertension in the Netherlands, de Kluizenaar et al. (2007) reported a correlation coefficient of 0.72 between modeled noise and modeled annual average PM10. In a study in Germany, Ising et al. (2004) reported a strong correlation (r = 0.84) between measurements of nighttime (0:00–6:00) traffic noise and 58–93 h measurements of NO2 at 25 locations. A recent study in Vancouver, BC, calculated correlations between 5-min noise and 2-week NO2 and NOx measured at 103 locations. They reported noise-NOx and noise-NO2 correlations of 0.64 and 0.53, respectively (Davies et al., 2009). A study in the Netherlands found a relatively poor correlation between yearly modeled noise and background black smoke (r = 0.24) (Beelen et al., 2008). Most recently, Selander et al. (2009) reported a correlation coefficient of 0.6 between long-term modeled estimates of Leq and NO2 in Sweden. Although these studies suggest the potential for confounding, all but the Vancouver work were conducted in Europe where differences in the vehicle fleet, roadway configuration, fuel composition, and urban design, as well as these studies’ frequent reliance on models, may limit the generalizabilty of the results to other settings. In summary, little is presently known about the spatial relationship between traffic-generated air pollution and noise in North America.

Here we present the results from a pilot investigation of the relationship between traffic-generated air pollution and noise in Chicago, IL, and Riverside County, CA. Our primary objective was to assess the potential for confounding in epidemiologic studies of chronic health effects by evaluating the correlations between noise and 3 markers of traffic-generated air pollution: NO, NO2, and ultrafine particles (UFPs). A secondary objective was to evaluate and compare the ability of simple roadway proximity metrics to predict measured levels of air pollution and noise.

2. Methods

2.1. Pollution measurements

This work leveraged off of the Multi-Ethnic Study of Atherosclerosis and Air Pollution (“MESA Air”). MESA Air is an ongoing investigation of chronic exposure to fine particulate matter (PM2.5) and other air pollutants in relation to the progression of subclinical atherosclerosis in 9 US communities. As part of its air pollution exposure assessment efforts, MESA Air collected simultaneous measurements of NO2 and NOx at up to 105 locations in each study community. Time-integrated NO2/NOx sampling was conducted over 2-week periods using passive Ogawa samplers attached to utility poles at approximately 2.5 m above the ground level. Three such 2-week sampling sessions were conducted in each study community to capture seasonal variations in spatial patterns.

MESA Air NO2/NOx sampling locations were selected to ensure spatial coverage around study participants and provide information in regions of high spatial variability of pollution, with emphasis on major roadways (i.e., those in census feature classification code [CFCC] categories A1, A2, and A3). Concentration gradients near roads were captured by deploying 6 samplers along a trajectory perpendicular to major “target” roadways. At each of these “gradient sites” 3 samplers were placed on both sides of the target roadway between 0 and 50, 50 and 100, and 100 and 350 m from the target roadway’s edge on small (CFCC category A4) streets or alleys. Samplers were also deployed in non-residential areas to capture the influence of multiple land uses, and additional samples (approximately 10% of the total) were collected at random locations within the study areas. NO concentrations were calculated by subtracting NO2 from NOx.

To assess the relationship between traffic-generated air pollution and noise we measured 5-min equivalent continuous sound pressure levels (Leq), reported in units of A-weighted decibels (dBA), at MESA Air NOx/NO2 sampling locations during 2 sampling sessions in Chicago, IL (December 5–8, 2006 and April 10–13, 2007) and 1 sampling session in Riverside, CA (April 17–19, 2007). Leq measurements in Chicago were made using a Larson Davis 870B sound level meter (SLM) (Larson Davis, Depew, New York), and in Riverside we used a Larson Davis 820 SLM. The SLMs were calibrated each morning prior to data collection. For sampling, the SLM was placed on a tripod approximately 1 m above ground level and as close as possible to the utility pole supporting the Ogawa sampler. Five-minute Leq measurements were collected under the strong assumption that these “grab samples” would be representative of long-term noise levels (Davies et al., 2009). During each 5-min Leq sample technicians recorded roadway characteristics and the presence of sporadic sources of noise (e.g., barking dogs, music, etc.). A single SLM was used for each sampling session and technicians moved from site to site, making a measurement at up to 20 locations per day. Noise sampling was conducted between 10:00 and 16:00 to avoid the influence of “rush hour” traffic and to maximize comparability between measurements. We did not sample during rain or snow due to concerns about the generalizability of noise levels measured under such conditions. There was some ice and snow on the ground in Chicago during the 2006 sampling session, and during all monitoring in both cities the field technicians recorded the road conditions (e.g., dry, wet, snow, ice, etc.) at the sampling location.

In addition to measuring noise, we also measured 5-min average UFP using the P-Trak (TSI, Shoreview, MN) during the 2007 sampling sessions in both cities. The P-Trak was zeroed daily prior to sampling. During sampling the instrument was placed on a tripod approximately 1 m above ground level and inside a sound-insulated toolbox near the SLM and the pole to which the Ogawa sampler was attached. The P-Trak has been validated based on laboratory comparisons with other condensation particle counters (CPCs), although a recent comparison in multiple microenvironments reported that the agreement between the P-Trak, which detects particles larger than approximately 30 nm, and a standard CPC depended on particle size and found that the P-Trak significantly underestimated UFP near a highway (Zhu et al., 2006b).

Repeated measurements of 5-min Leq and UFP were collected at selected locations to assess within-session and between-session variability in our measurements. Hourly wind speed at Chicago O’Hare and Riverside Community airports were obtained from the National Oceanic and Atmospheric Administration, and technicians recorded the wind direction during the 5-min noise and UFP measurements in both 2007 sampling sessions.

2.2. Data analysis

The representativeness of our 5-min measurements was assessed by evaluating the variability of measurements made at the same location at different times of day on different days within the same sampling session (both cities), and in different seasons (Chicago only). This was important since we collected 5-min noise measurements but our interest was in long-term levels. For our analyses, we computed averages of noise or UFP at locations with multiple measurements. In addition, to ensure comparability of results across the different pollutants, we included only locations with valid measures of Leq, NO, NO2, and UFP (if measured).

Because simple roadway proximity measures have become a common exposure surrogate in air pollution epidemiologic studies (Allen et al., 2009; Finkelstein et al., 2004; Hoek et al., 2002; Hoffmann et al., 2007, 2006; Maheswaran and Elliott, 2003; Tonne et al., 2007) we evaluated the relationships between our noise and air pollution measurements and roadway proximity variables (both binary and continuous) computed using the Dynamap 2000 road network (TeleAtlas, Lebanon, NH) in ArcGIS 9.2 (ESRI, Redlands, CA). First we calculated correlation coefficients between measured levels and the logarithm of the distance to 2 roadway categories: nearest major road (defined as CFCC classes A1, A2, and A3) and nearest highway (defined as CFCC classes A1 and A2). We also evaluated the relationships between our measured values and a binary indicator of roadway proximity defined as <100 m from a highway (A1 or A2) or <50 m from a major arterial (A3). This specific exposure surrogate has been used in at least 3 previous air pollution epidemiology studies (Allen et al., 2009; Finkelstein et al., 2004; Hoek et al., 2002). In addition to comparing the near roads and far from roads distributions, we also explored whether this binary variable was successful in stratifying the samples into high- and low-level groups, as defined by being above or below the median levels of noise and air pollution. We report the results by pollutant as the percent of observations correctly classified.

Finally, we calculated Pearson’s correlation coefficients between 5-min Leq and each of the traffic pollution indicators: NO (2-week average), NO2 (2-week average), and UFP (5-min average). In order to disentangle the regional (i.e., 10s to 100s of kilometers) and local-scale (i.e., 10s to 100s of meters) gradients of each pollutant, we first estimated the regional-scale gradients in air pollution and noise by conducting session-specific stepwise regressions (p<0.05 to enter and p<0.05 to remain in the model) of our measured values on variables expected to capture these large-scale gradients: distance to the city center in Chicago and location (i.e., distance north and distance east) within the study area in both cities. The variability from local sources was then estimated as the difference between our measurements and these modeled regional gradient surfaces. We calculated noise–air pollution correlation coefficients based on both the unadjusted measurements and the measurements after removing the influence of regional gradients. In addition, we calculated the correlation coefficients between noise and air pollution at gradient monitoring sites after stratifying measurements into locations upwind or downwind of the target roadway and less than or more than 100 m from the target roadway. For correlations involving NO or NO2, we categorized gradient monitoring locations into upwind or downwind based on the target roadway configuration and the dominant wind direction(s) measured at the airports over the 2-week sampling period (Fig. 1). The correlations between 5-min Leq and UFP measurements were categorized based on target roadway configuration and wind directions measured at O’Hare airport (2006 sampling session) or observed by field technicians during the 5-min measurements (2007 sampling sessions).

Fig. 1
Hourly wind roses during the three 2-week NO2/NOx sampling sessions.

3. Results

We sampled noise at 74 of 103 locations with NO2/NOx measurements during the first sampling session in Chicago (Table 1). Due to equipment problems and a snow storm the 2007 sampling session in Chicago included only 37 locations for Leq and 50 for UFP. In Riverside, we obtained valid noise and UFP measurements at 49 of 50 NO2/NOx measurement sites (fewer NO2/NOx samplers were deployed in Riverside because the MESA Air study area is smaller). Restricting our analyses to only those sites for which all measured pollutants were available resulted in 69 sites (covering approximately 2300 km2) for 2006 in Chicago, 36 (covering approximately 800 km2) for 2007 in Chicago, and 46 (covering approximately 400 km2) for Riverside.

Table 1
Summary statistics of sampling locations’ proximities to major roads and pollution measurements by city and sampling session.

As expected based on the roadway gradient site selection strategy in MESA Air, we captured a substantial amount of variability in NO, NO2, UFP, and Leq (Table 1). The range in Leq was approximately 25 dBA during all 3 monitoring sessions. There were some important differences in the monitoring locations between the 3 monitoring sessions. A lower proportion of measurements were made near highways (CFCC categories A1 or A2) during the first session in Chicago (10%) than during the 2007 sessions in Chicago (22%) or Riverside (26%) (Table 1). In addition, the proportion of gradient monitoring sites focused on highways was lower in Chicago (5 of 13 groups of samples in 2006 and 4 of 8 in 2007) than in Riverside (6 of 7). Average hourly wind speeds during noise measurements were higher in Chicago (6.4–6.7 m/s) than in Riverside (3.8 m/s) (Fig. 1).

We assessed temporal variation in our 5-min measurements and their ability to represent longer averaging times by comparing measurements made at different times of day within our 10:00–16:00 window (Table 2). Noise measurements were stable over time, with coefficients of variation (CV) at individual locations ranging between 0.5% and 12.9%. In contrast the UFP measurements were much more variable; the UFP CVs ranged between 58% and 67%, indicating that our 5-min UFP measurements do not provide a reliable estimate of longer-term concentrations.

Table 2
Summary statistics for repeated measurements made on different days and at different times of day during specific monitoring sessions.

We also evaluated the long-term variability in 5-min Leq measurements by comparing measurements from the December 2006 and April 2007 sampling sessions in Chicago. The 22 locations sampled during both sessions were highly correlated (r = 0.84; Fig. 2). The absolute value differences in repeat Leq measurements between the 2 seasons ranged between 0.3 and 9.0 dBA, with a median of 2.3 dBA. These results suggest that 5-min measurements made between 10:00 and 16:00 are stable across seasons and are representative of long-term average noise levels during those times of day. The between-season correlations for NO and NO2 at this subset of locations with noise measurements were moderate, with correlation coefficients of 0.59 and 0.42, respectively. Across all sites with NO/NO2 data in both seasons (N = 72) the between-season correlations for NO and NO2 were 0.65 and 0.56, respectively.

Fig. 2
Relationship between repeated 5-min Leq measurements in Chicago in December 2006 and April 2007.

Stepwise regression of our measured values on distance to city and location within the study area revealed some strong regional gradients, particularly for NO and NO2 (Table 3). In Chicago, combinations of these predictors explained 49–54% of the variability in NO and 70–82% of the variability in NO2. Distance to city alone explained 78% of the December, 2006 NO2 variance in Chicago (with decreasing concentrations at increasing distances from the city). The air pollution gradients were less pronounced in Riverside, although distance from the southern edge of the study area explained 17% and 27% of the variability in NO and NO2, respectively. As expected, our Leq and UFP measurements showed very weak regional-scale spatial patterns.

Table 3
Stepwise regression equations used to model regional gradients.

Our gradient sampling sites captured substantial variation in the measured pollutants in relation to distance from the “target” roadway (Fig. 3). When comparing regional gradient-adjusted concentrations <100 m vs. > 100 m from the target road, levels of NO and Leq were significantly (p<0.05) higher near the target road during all sessions, while NO2 was only elevated within 100 m during both of the 2007 sampling sessions (potentially due to the greater emphasis on highways in those sessions). UFP did not differ by road proximity, although this result is complicated by the temporal variation in UFP measurements described above. NO was most sensitive to wind direction, with significant differences by road proximity only on the downwind side, and consistently higher correlations with logarithmic distance to the target road on the downwind side.

Fig. 3
Relative concentrations adjusted for regional trends at monitoring sites upwind (white dots) and downwind (black dots) of targeted major roadway. Lines are logarithmic functions fit to the upwind and downwind data.

We evaluated the relationships between our measurements, adjusted for regional gradients to focus on local sources of variability, and simple surrogates of exposure based on roadway proximity (Table 4). Leq, NO, and NO2 were generally correlated with both distance metrics. The logarithm of distance to nearest major road was slightly better correlated with Leq (r = −0.41 to −0.52) than NO (r = −0.30 to −0.46) or NO2 (r = −0.22 to −0.40). Conversely, modest correlations with logarithmic distance to the nearest highway did not demonstrate consistent trends for NO (r = −0.43 to −0.53), NO2 (r = −0.39 to −0.59), or Leq (−0.18 to −0.57).

Table 4
Pearson’s correlation coefficients between pollutants and logarithm of roadway proximity after adjusting for regional gradients.

We also explored a binary roadway proximity variable (where “near roads” was defined as <100 m from an A1 or A2 road or <50 m from an A3 road) in relation to regional gradient-adjusted pollutant measurements. We compared the ability of this proximity variable to correctly identify levels of noise and air pollution by calculating the number of observations that were “correctly classified,” which was defined as measurements made near roads that were above the median level or measurements made distant from roads that were below the median. This binary roadway proximity variable consistently provided a better characterization of noise (65–87% of measurements classified correctly) than of NO (59–72%), NO2 (61–70%), or UFP (44–61%).

The overall correlations between measured Leq and traffic-generated air pollution were moderate during both of the 2007 sampling sessions (left half of Table 5), with correlation coefficients of 0.40–0.60 for NO and 0.38–0.46 for NO2 (all p<0.05). In contrast, the correlations during the 2006 sampling session in Chicago were weak for NO (r = 0.20) and NO2 (r = −0.08). After removing the influence of the regional variation primarily in NO and NO2, correlations between noise and air pollution generally increased (right half of Table 5). The most dramatic increase was in the Leq–NO correlation for 2006 in Chicago, which increased from 0.20 (p<0.10) to 0.49 (p<0.01). Wind speed was correlated with noise only during the December 2006 monitoring session in Chicago, although adjusting for wind speed did not influence the noise–air pollution correlations. Other noise sources recorded by technicians were not associated with measured noise levels (results not shown).

Table 5
Pearson’s correlation coefficients between Leq and air pollutants.

Finally, we further examined the noise–air pollution correlations by evaluating the impact of wind direction and road proximity on the measurements made at roadway gradient monitoring sites (Table 6). The Leq–NO correlations were consistently higher downwind of major target roads (r = 0.53–0.74; all p<0.05) than they were upwind (r = 0.35–0.65). Although there was not a significant relationship between Leq and NO2 in either wind direction during the 2006 sampling session, the Leq–NO2 correlations during both 2007 sampling sessions were also higher downwind (r = 0.57 and 0.71) than upwind (r = 0.37 and 0.60). The effect of road proximity was not consistent across cities. In Chicago, higher correlations were noted within 100 m of the target road, while in Riverside we generally found higher correlations for sites greater than 100 m from the target road.

Table 6
Pearson’s correlation coefficients at roadway gradient monitoring sites after adjusting for regional gradients. The number of measurements used to calculate the correlation is in parentheses.

4. Discussion

To our knowledge, this is the first investigation of the relationship between traffic-related air pollution and noise in the US. The temporal variability of noise was found to be much lower than that of NO or NO2 in Chicago, perhaps due to the greater impact of meteorology on air pollution concentrations. In fact, 5-min grab samples of noise repeated at the same location were found to be quite stable over time (between-season r = 0.84). This stability was extremely important for this study since we used single instruments to obtain 5-min grab samples due to the costs and security concerns associated with deploying multiple monitors. Furthermore, these results suggest that short-term noise measurements may provide a useful indicator of long-term averages of community noise in the absence of extensive noise monitoring or data-intensive noise models. Alberola and colleagues (2005) analyzed hourly noise measurements (6:00–19:00) collected over 2 weeks at 50 locations impacted primarily by road noise and found that variability in hourly measurements increased with decreasing noise level. This suggests that our 5-min measurements may have been most representative of longer-term averages at the loudest locations, i.e., near major roads. In contrast to our noise measurements, we did not observe temporal stability for UFP, which demonstrated large differences in concentrations across time. This instability is likely due to the fact that as primary pollutants with short atmospheric lifetimes, UFPs are highly sensitive to the type of vehicles and the local wind conditions (Zhu et al., 2006a). Because of this variability and the fact that 10–26% of our measurements were made within 100 m of a highway, where the P-Trak has been shown to significantly underestimate UFP concentrations, we place greatest emphasis on the results for the other traffic-related air pollutants, NO and NO2.

Logarithmic distances to nearby major roads were only moderately (r≈0.4–0.6) correlated with Leq, NO, and NO2. These correlations are lower than in previous studies of the relationship between NO2 and logarithmic distance to roads, which have reported correlations of 0.83 in the Netherlands (Roorda-Knape et al., 1998), 0.94 in Montreal (Gilbert et al., 2003), and 0.97 in Sweden (Pleijel et al., 2004). One possible explanation for this discrepancy is that our sampling was conducted on a variety of target roads, including major arterials, to estimate concentrations in areas near MESA Air residences, while previous studies focused primarily on highways. The fact that simple proximity measures were found to predict noise at least as well as air pollution suggests that these exposure surrogates are not a unique identifier in epidemiologic studies of health outcomes for which noise and air pollution are both hypothesized causes.

NO and NO2 were moderately correlated with noise levels in both cities. These correlations were generally similar to the noise-NO2 and/or noise-NOx correlations reported in Oslo, Stockholm, and Vancouver (Davies et al., 2009; Klaeboe et al., 2000; Selander et al., 2009), but lower than the noise-PM10 (r = 0.72) and noise-NO2 (r = 0.84) correlations reported in the Netherlands and Germany, respectively (de Kluizenaar et al., 2007; Ising et al., 2004).

Since both NO and NO2 had clear gradients across the study areas (noise and UFP had very weak regional-scale spatial patterns), these larger trends were found to impact our noise–air pollution correlations. Large-scale pollution gradients may also explain the relatively low correlations between noise and background black smoke (r = 0.24) observed in a recent study in the Netherlands (Beelen et al., 2008). These findings imply that epidemiologic investigations exploiting the influence of local roadways might be more vulnerable to confounding than studies focused on larger regional differences. However, the importance of this finding is difficult to quantify as the relative importance of localized vs. regional pollution gradients has not yet been determined. Numerous studies have reported adverse cardiovascular health effects among individuals residing in close proximity to major roads (Finkelstein et al., 2004; Hoek et al., 2002; Hoffmann et al., 2007, 2006; Maheswaran and Elliott, 2003; Tonne et al., 2007), suggesting an important role for more localized pollution gradients, although 2 recent studies have reported elevated risks in relation to urban scale fine particulate matter air pollution gradients across the Los Angeles urban area (Jerrett et al., 2005b; Kunzli et al., 2005).

Wind direction was an important modifier of the relationships between air pollution and noise. NO concentrations and distance decay relationships were most sensitive to wind direction. In contrast, noise had similar distance decay relationships upwind, and the similarity of 5-min noise measurements made at 22 locations in different seasons (with different wind characteristics) provides further evidence that noise was minimally impacted by wind direction. These differential wind direction effects impacted the noise–air pollution correlations; the strongest, most consistent correlations were observed between noise and NO downwind of major roads (r = 0.53–0.74).

Some important limitations of this study should be considered. First, our noise (5-min) and NO/NO2 (2-week) measurements were conducted over different time periods. Although we cannot determine the impact of this difference on the observed correlations, one might speculate that measurements made over the same duration would be better correlated than those with significantly different durations, suggesting that the correlations presented here are underestimates. In addition, since our “grab sample” approach to characterizing spatial patterns also involved a temporal component, we limited our noise measurements to non-rush hour periods to minimize the influence of temporal differences on our assessment of spatial patterns. However, this approach may have also biased the observed correlations with air pollution, or may not represent the most biologically relevant correlations if nighttime noise exposures are most detrimental to health (Babisch, 2006). An additional limitation to this study design was our dependence on wind speed and direction measured at a single location in each community.

While the primary focus of this paper has been confounding in cardiovascular epidemiology, consideration of the interplay between noise and air pollution may also be important for respiratory health research. Several investigations have reported associations between traffic-generated air pollution and the development and exacerbation of asthma (Brauer et al., 2007; Gauderman et al., 2005; Gordian et al., 2006), and recent research suggests that stress may modify these effects (Chen et al., 2008; Clougherty et al., 2007). Noise is thought to act on cardiovascular health through repeated noise-induced stress responses (Babisch, 2002; Maschke et al., 2000) suggesting the potential need to also assess noise exposure in studies of traffic-related air pollution and asthma.

In conclusion, moderate correlations suggest the potential for confounded results if both noise and air pollution are not accurately assessed in epidemiological studies of traffic and health. Although very few epidemiologic studies have included both air pollution and noise in health effects models (Beelen et al., 2008; de Kluizenaar et al., 2007; Schwela et al., 2005; Selander et al., 2009), imperfect correlations between these exposures present opportunities for disentangling their impacts on health, and methods for analyzing correlated environmental exposures in health effects studies continue to emerge (Dominici et al., 2008; MacLehose et al., 2007; Thomas, 2007). Future studies of endpoints such as MI, for which there are hypothesized physiological mechanisms and preliminary epidemiological evidence implicating both noise (Babisch et al., 2005; Selander et al., 2009) and air pollution (Tonne et al., 2007), may require more sophisticated exposure assessments involving measurements and/or models of both pollutants. Consideration of prevailing wind direction(s) and/or regional-scale air pollution gradients may allow investigators to minimize the potential for confounding.


Funding for this work was provided by the Simon Fraser University President’s Research Grant. The MESA Air study is funded by the US Environmental Protection Agency (Grant R831697). We thank Dr. Bruce Allen for assistance with equipment preparation, Melissa Symon for help with data collection, Anne Ho and Jim Sullivan for data processing, and Amber Pearson for calculation of geographical variables. Mention of trade names or commercial products does not constitute an endorsement or recommendation for use.


[star]Funding Sources: Funding for this work was provided by the Simon Fraser University President’s Research Grant and the US Environmental Protection Agency (Grant R831697). Joel Kaufman was supported by the National Institute of Environmental Health Sciences through grant K24ES013195.


  • Adar SD, Kaufman JD. Cardiovascular disease and air pollutants: evaluating and improving epidemiological data implicating traffic exposure. Inhalation Toxicology. 2007;19:135–149. [PubMed]
  • Ainslie B, Steyn DG, Su J, Buzzelli M, Brauer M, Larson T, Rucker M. A source area model incorporating simplified atmospheric dispersion and advection at fine scale for population air pollutant exposure assessment. Atmospheric Environment. 2008;42(10):2394–2404.
  • Alberola J, Flindell IH, Bullmore AJ. Variability in road traffic noise levels. Applied Acoustics. 2005;66(10):1180–1195.
  • Allen R, Criqui MH, Diez Roux AV, Allison M, Shea S, Detrano R, Sheppard L, Wong N, Hinckley Stukovsky K, Kaufman JD. Fine particulate air pollution, proximity to traffic, and aortic atherosclerosis. Epidemiology. 2009 in press. [PMC free article] [PubMed]
  • Babisch W. The noise/stress concept, risk assessment and research needs. Noise & Health. 2002;4(16):1–11. [PubMed]
  • Babisch W. Transportation noise and cardiovascular risk: updated review and synthesis of epidemiological studies indicate that evidence has increased. Noise & Health. 2006;8:1–29. [PubMed]
  • Babisch WF, Beule B, Schust M, Kersten N, Ising H. Traffic noise and risk of myocardial infarction. Epidemiology. 2005;16(1):33–40. [PubMed]
  • Beckerman B, Jerrett M, Brook JR, Verma DK, Arain MA, Finkelstein MM. Correlation of nitrogen dioxide with other traffic pollutants near a major expressway. Atmospheric Environment. 2008;42(2):275–290.
  • Beelen R, Hoek G, Houthuijs D, van den Brandt PA, Goldbohm RA, Fischer P, Schouten LJ, Armstrong B, Brunekreef B. The joint association of air pollution and noise from road traffic with cardiovascular mortality in a cohort study. Occupational and Environmental Medicine. 2008 available online. [PubMed]
  • Bluhm GL, Berglind N, Nordling E, Rosenlund M. Road traffic noise and hypertension. Occupational and Environmental Medicine. 2007;64(2):122–126. [PMC free article] [PubMed]
  • Brauer M, Hoek G, Smit HA, de Jongste JC, Gerritsen J, Postma DS, Kerkhof M, Brunekreef B. Air pollution and development of asthma, allergy and infections in a birth cohort. European Respiratory Journal. 2007;29(5):879–888. [PubMed]
  • Calixto A, Diniz FB, Zannin PHT. The statistical modeling of road traffic noise in an urban setting. Cities. 2003;20(1):23–29.
  • Chen E, Schreier HMC, Strunk RC, Brauer M. Chronic traffic-related air pollution and stress interact to predict biological and clinical outcomes in asthma. Environmental Health Perspectives. 2008;116(7):970–975. [PMC free article] [PubMed]
  • Clougherty JE, Levy JI, Kubzansky LD, Ryan PB, Suglia SF, Canner MJ, Wright RJ. Synergistic effects of traffic-related air pollution and exposure to violence on urban asthma etiology. Environmental Health Perspectives. 2007;115(8):1140–1146. [PMC free article] [PubMed]
  • Davies HW, Vlaanderen J, Henderson Sl, Brauer M. Correlation between co-exposures to noise and air pollution from traffic sources. Occupational and Environmental Medicine. 2009 in press. [PubMed]
  • de Kluizenaar Y, Gansevoort RT, Miedema HME, de Jong PE. Hypertension and road traffic noise exposure. Journal of Occupational and Environmental Medicine. 2007;49(5):484–492. [PubMed]
  • Dominici F, Wang C, Crainiceanu C, Parmigiani G. Model selection and health effect estimation in environmental epidemiology. Epidemiology. 2008;19(4):558–560. [PubMed]
  • Finkelstein MM, Jerrett M, Sears MR. Traffic air pollution and mortality rate advancement periods. American Journal of Epidemiology. 2004;160(2):173–177. [PubMed]
  • Gauderman WJ, Avol E, Lurmann F, Kuenzli N, Gilliland F, Peters J, McConnell R. Childhood asthma and exposure to traffic and nitrogen dioxide. Epidemiology. 2005;16(6):737–743. [PubMed]
  • Gilbert NL, Goldberg MS, Brook JR, Jerrett M. The influence of highway traffic on ambient nitrogen dioxide concentrations beyond the immediate vicinity of highways. Atmospheric Environment. 2007;41(12):2670–2673.
  • Gilbert NL, Woodhouse S, Stieb DM, Brook JR. Ambient nitrogen dioxide and distance from a major highway. Science of the Total Environment. 2003;312(1–3):43–46. [PubMed]
  • Gordian ME, Haneuse S, Wakefield J. An investigation of the association between traffic exposure and the diagnosis of asthma in children. Journal of Exposure Science and Environmental Epidemiology. 2006;16(1):49–55. [PubMed]
  • Hoek G, Brunekreef B, Goldbohm S, Fischer P, van den Brandt PA. Association between mortality and indicators of traffic-related air pollution in the Netherlands: a cohort study. Lancet. 2002;360(9341):1203–1209. [PubMed]
  • Hoffmann B, Moebus S, Mohlenkamp S, Stang A, Lehmann N, Dragano N, Schmermund A, Memmesheimer M, Mann K, Erbel R, Jockel KH. Residential exposure to traffic is associated with coronary atherosclerosis. Circulation. 2007;116(5):489–496. [PubMed]
  • Hoffmann B, Moebus S, Stang A, Beck EM, Dragano N, Mohlenkamp S, Schmermund A, Memmesheimer M, Mann K, Erbel R, Jockel KH. Residence close to high traffic and prevalence of coronary heart disease. European Heart Journal. 2006;27(22):2696–2702. [PubMed]
  • Hothersall DC, Chandler-Wilde SN. Prediction of the attenuation of road traffic noise with distance. Journal of Sound and Vibration. 1987;115(3):459–472.
  • Ising H, Lange-Asschenfeldt H, Moriske H-J, Born J, Eilts M. Low Frequency noise and stress: bronchitis and cortisol in children exposed chronically to traffic noise and exhaust fumes. Noise & Health. 2004;6(23):23–30. [PubMed]
  • Jerrett M, Arain A, Kanaroglou P, Beckerman B, Potoglou D, Sahsuvaroglu T, Morrison J, Giovis C. A review and evaluation of intraurban air pollution exposure models. Journal of Exposure Analysis and Environmental Epidemiology. 2005a;15(2):185–204. [PubMed]
  • Jerrett M, Burnett RT, Ma RJ, Pope CA, Krewski D, Newbold KB, Thurston G, Shi YL, Finkelstein N, Calle EE, Thun MJ. Spatial analysis of air pollution and mortality in Los Angeles. Epidemiology. 2005b;16(6):727–736. [PubMed]
  • Klaeboe R, Kolbenstvedt M, Clench-Aas J, Bartonova A. Oslo traffic study—part 1: An integrated approach to assess the combined effects of noise and air pollution on annoyance. Atmospheric Environment. 2000;34(27):4727–4736.
  • Kunzli N, Jerrett M, Mack WJ, Beckerman B, LaBree L, Gilliland F, Thomas D, Peters J, Hodis HN. Ambient air pollution and atherosclerosis in Los Angeles. Environmental Health Perspectives. 2005;113(2):201–206. [PMC free article] [PubMed]
  • MacLehose RF, Dunson DB, Herring AH, Hoppin JA. Bayesian methods for highly correlated exposure data. Epidemiology. 2007;18(2):199–207. [PubMed]
  • Maheswaran R, Elliott P. Stroke mortality associated with living near main roads in England and Wales—a geographical study. Stroke. 2003;34(12):2776–2780. [PubMed]
  • Maschke C, Rupp T, Hecht K. The influence of stressors on biochemical reactions-a review of present scientific findings with noise. International Journal of Hygiene and Environmental Health. 2000;203(1):45–53. [PubMed]
  • Pleijel H, Karlsson GP, Gerdin EB. On the logarithmic relationship between NO2 concentration and the distance from a highroad. Science of the Total Environment. 2004;332(1–3):261–264. [PubMed]
  • Roorda-Knape MC, Janssen NAH, de Hartog JJ, van Vliet PHN, Harssema H, Brunekreef B. Air pollution from traffic in city districts near major motorways. Atmospheric Environment. 1998;32(11):1921–1930.
  • Schwela D, Kephalopoulos S, Prasher D. Confounding or aggravating factors in noise-induced health effects: air pollutants and other stressors. Noise & Health. 2005;7(28):41–50. [PubMed]
  • Selander J, Nilsson ME, Bluhm G, Rosenlund M, Lindqvist M, Nise G, Pershagen G. Long-term exposure to road traffic noise and myocardial infarction. Epidemiology. 2009 in press. [PubMed]
  • Su JG, Brauer M, Ainslie B, Steyn D, Larson T, Buzzelli M. An innovative land use regression model incorporating meteorology for exposure analysis. Science of the Total Environment. 2008;390(2–3):520–529. [PubMed]
  • Thomas DC. Dissecting effects of complex mixtures—who’s afraid of informative priors? Epidemiology. 2007;18(2):186–190. [PubMed]
  • Tobias A, Diaz J, Saez M, Alberdi JC. Use of Poisson regression and Box–Jenkins models to evaluate the short-term effects of environmental noise levels on daily emergency admissions in Madrid, Spain. European Journal of Epidemiology. 2001;17(8):765–771. [PubMed]
  • Tonne C, Melly S, Mittleman M, Coull B, Goldberg R, Schwartz J. A case–control analysis of exposure to traffic and acute myocardial infarction. Environmental Health Perspectives. 2007;115(1):53–57. [PMC free article] [PubMed]
  • van Kempen E, Kruize H, Boshuizen HC, Ameling CB, Staatsen BAM, de Hollander AEM. The association between noise exposure and blood pressure and ischemic heart disease: a meta-analysis. Environmental Health Perspectives. 2002;110(3):307–317. [PMC free article] [PubMed]
  • Zhu YF, Hinds WC, Kim S, Sioutas C. Concentration and size distribution of ultrafine particles near a major highway. Journal of the Air & Waste Management Association. 2002;52(9):1032–1042. [PubMed]
  • Zhu YF, Kuhn T, Mayo P, Hinds WC. Comparison of daytime and nighttime concentration profiles and size distributions of ultrafine particles near a major highway. Environmental Science & Technology. 2006a;40(8):2531–2536. [PubMed]
  • Zhu YF, Yu N, Kuhn T, Hinds WC. Field comparison of P-trak and condensation particle counters. Aerosol Science and Technology. 2006b;40(6):422–430.
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...