Combined impact of dynamic air pollution and transportation noise on cardiometabolic disorders in San Diego County

Abstract

Air pollution and noise exposure may synergistically contribute to increased cardiometabolic disorders; however, few studies have examined this potential interaction or considered exposures beyond residential location. The combined impact of dynamic air pollution and transportation noise on cardiometabolic disorders in San Diego County was investigated. Using data from the Community of Mine Study (2014-2017), 602 ethnically diverse participants were assessed for obesity, dyslipidemia, hypertension, and metabolic syndrome (MetS), using anthropometric measurements and biomarkers from blood samples. Time-weighted measures of exposure to fine particulate matter, nitrogen dioxide (NO₂), and road and aircraft noise were calculated using global positioning system mobility data and kernel density estimation. Generalized estimating equation models were used to analyze associations. Interactions were assessed on the multiplicative and additive scales using relative excess risk due to interaction (RERI). We found that air pollution and noise interact to affect metabolic disorders on both multiplicative and additive scales. The effect of noise on obesity and MetS was higher when air pollution level was higher. The RERIs of aircraft noise and NO₂ on obesity and MetS were 0.13 (95% CI, 0.03-0.22) and 0.13 (95% CI, 0.02-0.25), respectively. These findings suggest aircraft noise and air pollution may have synergistic effects on obesity and MetS.

This article is part of a Special Collection on Environmental Epidemiology.

Introduction

Exposure to pollution is the leading environmental risk factor for noncommunicable diseases, including cardiovascular diseases (CVDs).¹ An estimated 47 million people in the United States are living with cardiometabolic disorders,² including dyslipidemia, hypertension, metabolic syndrome (MetS), and obesity, which are major risk factors for CVDs and metabolic diseases.^3,4 Among the modifiable risk factors of cardiometabolic disorders, environmental pollution, including air pollution and transportation noise, has received increasing attention in recent decades.^5-7 Both air pollution and noise exposure have separately been shown to affect metabolic disorders. Evidence linking air pollution and the risk of cardiometabolic disorders such as hypertension,^5,8-10 diabetes,¹¹ and obesity¹² is increasing.¹³ Similarly, the evidence linking transportation noise with metabolic disorders has increased.^14,15 Although findings for noise exposure have been scarcer and more mixed than results for air pollution, transportation noise was associated with obesity,^16,17 hypertension,^18-20 increased risk of diabetes,^21-23 hyperglycemia,²⁴ and MetS.^25,26

Most of the epidemiologic evidence regarding air pollution and noise exposure is focused on single exposures, disregarding potential joint effects, although both exposures can share a common source (eg, transportation) and some similar mechanistic pathways (eg, oxidative stress, vascular dysfunction, autonomic imbalance, metabolic abnormalities).²⁷ Therefore, they may act synergistically on cardiometabolic health. In previous studies in which both exposures were considered, their potential interaction rarely was examined.²⁸ This can lead to misestimation of singular and joint associations. Few studies have investigated the interaction between noise and air pollution on CVDs^29,30 such as coronary heart disease,⁶ myocardial infarction,⁷ stroke,³¹ and heart failure.³² Overall, among these studies, the greatest effects were found for the combination of high exposure to both noise and air pollution, even though these were not always statistically significant. A 2023 systematic review indicated the need for more studies investigating interactions between noise and air pollution on cardiovascular outcomes.²⁸ Because these exposures are likely to occur together, understanding if their joint effects on cardiometabolic health are more pronounced than single effects may help inform future intervention efforts for reduction of both pollutants.

Numerous studies have investigated how noise impacts may be confounded by air pollution,^17,28 or vice versa.^20,33 However, to our knowledge, only 2 studies have investigated interactions between noise and air pollution and metabolic disorders, and the interactions were only assessed on the multiplicative scale. In a Dutch national health survey of 387 195 adults, Klompmaker et al.³⁴ found a weak indication for potential effect measure modification between air pollution and road traffic noise. The association of air pollution with the odds of diabetes was slightly stronger in the highest road traffic noise quintile compared with the other quintiles, although it was not statistically significant. In a study using the UK Biobank (n > 500 000 participants), Huang et al.³⁵ found that the effect of road traffic noise on hypertension incidence was exacerbated in the presence of higher air pollution levels.

A weakness of previous studies was the use of static exposure,^16,34,36-45 or the assumption that the participant’s exposure is represented by exposure measures within a radius of their residence. This can be far from reality.⁴⁶ Dynamic exposure with time-weighted spatial averaging approaches uses the participant’s activity space estimated by global positioning system (GPS) data. Because a person’s exposure to air pollution and noise can vary significantly depending on their location and the time spent in these locations, dynamic exposure reflects real-world conditions and thus better captures the variations in exposure that individuals experience in their daily lives,⁴⁷ which may more accurately capture the relationship between exposure to pollutants and health.

The main goal of this study is to understand the combined effect of air pollution and transportation noise on the risk of cardiometabolic disorders in an ethnically diverse population in San Diego County, CA, by analyzing interactions at multiplicative and additive scales.

Methods

Study population

The Community of Mine Study is a cross-sectional observational study conducted from 2014 to 2017 in San Diego County, CA.⁴⁸ A total of 602 adults aged 35-80 years who had been living at their current address for at least 6 months were randomly selected from census block groups with a range of neighborhood walkability. Participants attended a clinic where blood pressure, height, weight, and hip and waist circumference were recorded, and blood and urine samples were collected. Total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and fasting glucose levels were measured in fasting blood samples. Additionally, participants were asked to complete a self-reported survey to collect demographic characteristics (eg, age, sex, ethnicity, income, employment status) and to wear a GPS and accelerometer sensor for 14 days. The main objective of the study was to advance methods of cancer risk exposure assessment by measuring both neighborhood access and total exposure to healthy environments by integrating GPS data. Full protocol and inclusion criteria are described in the report of Jankowska et al.⁴⁸

Outcomes

This study focused on 4 cardiometabolic disorder outcomes: dyslipidemia, hypertension, MetS, and obesity. Dyslipidemia was defined as the presence of 1 or more of the following conditions: total cholesterol ≥200 mg dL^-1, triglycerides ≥150 mg dL^-1, low-density lipoprotein cholesterol ≥130 mg dL^-1, or HDL cholesterol < 40 mg dL^-1. Hypertension was characterized as blood pressure of at least 140/90 mm Hg. MetS was defined as the presence of at least 3 of the following risk determinants: increased waist circumference (>102 cm [>40 in] for men, > 88 cm [>35 in] for women); elevated triglyceride levels (≥150 mg dL^-1); low HDL cholesterol (<40 mg dL^-1 in men, < 50 mg dL^-1 in women); hypertension (≥130/≥80 mm Hg); and impaired fasting glucose level (≥110 mg dL^-1). Obesity was defined as body mass index (BMI) > 30 kg m^-2.

Covariates

Covariates comprised age, sex, ethnicity (Hispanic/Latino vs non-Hispanic), annual income (3 categories: < $30 000, $30 000-$55 000, and > $55 000); current smoking status (yes or no); BMI (calculated as weight [kg] divided by the square of height [m²] and categorized as lean/normal [<25 kg m^-2], overweight [25-30 kg m^-2], and obese [>30 kg m^-2]); the Normalized Difference Vegetation Index (time-weighted measures of greenness exposure using GPS data); and neighborhood economic level (calculated from the California Health Place Index, taking into account neighborhood poverty, employment, and income level). Covariates were chosen a priori using a directed acyclic graph.

Environmental exposures

Air pollution

The modeling of air pollution was done based on daily observations of fine particulate matter (PM_2.5) and nitrogen dioxide (NO₂; μg m^-3) from 8 Environmental Protection Agency Air Quality System monitoring stations in San Diego County. Spatial–temporal interpolation of PM_2.5 and NO₂ concentrations was assessed with an inverse distance weighting approach to calculate the yearly average of air pollution from 2014 to 2017. We obtained a prediction R² of 0.80 and 0.78 for PM_2.5 and NO₂, respectively (Figure S1). The year a participant completed the study was matched to the same year of air pollution. If data from the year a participant completed the study were missing, air pollution data from the next subsequent year were used. We estimated annual exposure to each pollutant at the zip code population-weighted centroid. Overall, we obtained exposure to both NO₂ and PM_2.5 for all zip codes within our study area.

Transportation noise

Transportation noise exposures were assessed by the US Department of Transportation.⁴⁹ Annual average noise levels from aircraft and road were estimated for the year 2016. The national transportation noise inventory was developed using a 24-hour equivalent sound level noise metric. The noise map was generated by implementing the Aviation Environmental Design Tool, version 2b, developed by the Federal Aviation Administration, and the acoustic algorithms from the Traffic Noise Model 2.5 (TNM) proposed by the Federal Highway Administration. The transportation noise map was generated using traffic data, roadway inventory, and aircraft flight operation data, along with simplifying assumptions; namely, atmospheric absorption for aviation noise, nonhomogenous atmospheric effects in road traffic noise modeling, the TNM’s default temperature and humidity levels (68 °F [20 °C], 50% relative humidity), acoustically soft ground, average pavement material and texture of the road, and distribution of average annual daily traffic data across 24 hours. The modeling engines of the National Transportation Noise Mapping Tool ; the Aviation Environmental Design Tool, version 2b; and the Federal Highway Administration’s TNM models were validated previously. Spatial distribution of aircraft and road noise in 2016 across San Diego County is shown in Figure S2.

Dynamic exposures

Kernel density estimates were used to estimate the daily activity space of participants by combining all valid daily GPS points using a 200 m bandwidth and quartic kernel function. Valid GPS points were defined as having latitude and longitude coordinate data and were coded as “wearing the device” from the corresponding minute-level accelerometer data as calculated using the algorithm of Choi et al.⁵⁰ Applying map algebra to activity space and the 4 environmental layers (ie, PM_2.5, NO_2, aircraft noise, and road noise) resulted in daily time-weighted mean environmental exposure to air pollution and transportation noise for participants. The daily exposure values were averaged for each participant over the study period and transformed into z scores. Dynamic exposure calculation was implemented in Python, version 3.7, with the ArcPy package.

Static exposures

Static measures of noise exposure were assessed by calculating the zonal mean noise exposure by census tract. Each exposure was then assigned to the participant’s census tract based on their home address. Static measures of air pollution were assessed by calculating circular buffers of 1600 m around the participant’s address.

Statistical analysis

Pearson’s correlations between the exposure variables, greenness, and neighborhood economic level were calculated.

Single- and multi-exposure models

Generalized estimating equation models with exchangeable matrixes were applied, accounting for participants clustering within a zip code, to assess whether the exposure variables (ie, noise and air pollution) were associated with cardiometabolic disorders (namely, dyslipidemia, hypertension, obesity, and MetS). Modified Poisson regressions were used to account for the high prevalence of the binary outcomes of interest.^51,52 Four distinct models were applied with increasing degrees of covariate adjustment. Model 1 was a crude model. Model 2 was adjusted for age, sex, ethnicity, and income. Model 3 was additionally adjusted for greenness and neighborhood economic level. The final model, model 4, was additionally adjusted for air pollution (or transportation noise). For each primary exposure variable, a linear term, and natural splines (1, 2, 3, and 5 knots) were used to determine the linearity of the exposure-response relationship. To test whether the goodness of fit of the models with splines was significantly better than the goodness of fit of the linear models, the correlation information criterion was computed.⁵³ In supplementary analyses, we used generalized estimating equation models with static noise and air pollution exposures.

Interaction analysis

To evaluate whether the effect of transportation noise on cardiometabolic disorders was modified by air pollution, we assessed interaction at both multiplicative and additive scales. Interaction on a multiplicative scale means that the combined effect is different than the product of the individual effects. Interaction on an additive scale means that the combined effect of 2 exposures is different (larger or smaller) than the sum of the individual effects of the 2 exposures.⁵⁴ Interaction at the multiplicative scale was assessed by including an interaction term (between both exposures) in the models. Additionally, the associations were investigated across strata: individuals with air pollution exposures (PM_2.5 or NO₂) below the median level vs individuals exposed to air pollution at levels equal to or higher than the median exposure level. We performed a Cochran Q test to determine if regression coefficients between transportation noise and outcomes differed by air pollution level. Interaction on the additive scale was performed by computing relative excess risk due to interaction (RERI) with continuous exposure variables on the fully adjusted model (model 4). Relative excess risk due to interaction is defined as follows: RERI = RR_AP + N+ − RR_AP + N− − RR_AP − N+ + 1, where air pollution is denoted as AP and transportation noise as N; RR_AP + N+ corresponds to the relative risk (RR) among participants exposed to high levels of air pollution and noise; RR_AP + N- corresponds to the RR among participants exposed to high levels of air pollution and low levels of noise; and RR_AP − N+ corresponds to the RR among participants exposed to low levels of air pollution and high levels of noise. The direction of the RERI is needed to understand the public health relevance of interaction and can be used to identify which subgroups would benefit most from intervention.⁵⁴ An RERI > 0 implies there is a positive interaction (ie, air pollution could exacerbate noise effects and thus a public health intervention would have a larger impact on groups exposed to both pollutions), whereas RERI < 0 implies a negative interaction. All analyses were performed with R, version 4.2.1.

Results

Study population

Among the 602 participants in the Community of Mine Study, 6 participants had no data for dyslipidemia and obesity. Additionally, 21 participants had missing data for covariates and were excluded from this analysis. Overall, 575 participants were included in the population sample, and 560 participants were included for analysis with PM_2.5 exposure (because 15 participants had no dynamic PM_2.5 exposure) (see the flow chart in Figure S3). Among these 575 participants, 55.6% were women (n = 335), the median (IQR) age was 60 (50, 68) years, and 42% (n = 252) self-identified as Hispanic/Latino (Table 1). Few participants were smokers (7.6%). The median BMI was 27.8, which is considered overweight. The median time spent at the current address was 15 years, with a minimum value of 7 years.

Table 1.

Distribution of participant characteristics from the Community of Mine Study (n = 575)

Characteristic	No. (%) a
Female sex	320 (55.6)
Age, mean (SD), years	59.9 (10.8)
Ethnicity
Non-Hispanic	336 (58.4)
Hispanic/Latino	239 (41.6)
Race
American Indian or Alaska Native	24 (4.2)
Asian	21 (3.7)
Black or African American	17 (3.0)
Native Hawaiian or Other Pacific Islander	8 (1.4)
White	404 (70.3)
Unknown or other race/ethnicity	101 (17.6)
Smoking	43 (7.5)
Presence of metabolic disorders
Dyslipidemiab	273 (47.5)
Hypertensionc	138 (24)
Obesityd	195 (33.9)
Metabolic syndromee	114 (19.8)
Contextual indicators, median (IQR)
Years at current address	15 (7, 25)
Greenness (NDVI)	6.5-4 (4.5-4, 9.5-4)
Neighborhood economicf	0.21 (-0.49, 0.80)
Neighborhood educationg	0.09 (-0.32, 0.86)
Neighborhood health careh	0.26 (-0.50, 0.83)
Neighborhood incomei	63 848 (49 189, 87 333)

^aData reported as no. (%) unless otherwise indicated.

^bDyslipidemia was defined as the presence of ≥ 1 of the following conditions: total cholesterol ≥ 200 mg dL^-1; triglycerides ≥ 150 mg dL^-1; low-density lipoprotein cholesterol ≥ 130 mg dL^-1; high-density lipoprotein (HDL) cholesterol < 40 mg dL^-1.

^cHypertension was defined as blood pressure ≥ 130 mm Hg systolic or ≥ 80 mm Hg diastolic.

^dObesity was defined as body mass index ≥ 30 kg m^-2.

^eMetabolic syndrome was defined as the presence of ≥ 3 of the following: waist circumference > 102 cm for men, > 88 cm for women; triglycerides ≥ 150 mg dL^-1; HDL cholesterol < 40 mg dL^-1 in men, < 50 mg dL^-1 in women; hypertension ≥ 130/≥ 80 mm Hg; and fasting glucose ≥ 110 mg dL^-1.

^fNeighborhood economic was defined as neighborhood poverty, employment, and income level (from the California Health Place Index [HPI]).

^gNeighborhood education was defined as neighborhood school enrollment level (from the HPI index).

^hNeighborhood health care was defined as the number of insured adults at census-tract level.

ⁱNeighborhood income was defined as neighborhood income level (median household income).

Abbreviation: NDVI, Normalized Difference Vegetation Index (dynamic exposure).

Almost half of the population (48%; n = 289) had dyslipidemia, 24% (n = 143) had hypertension, 20% (n = 119) had MetS, and 35% (n = 208) were obese. Participants’ dynamic exposures are detailed in Table S1. We found a strong correlation between PM_2.5 and NO₂ (Pearson test P = 0.8), but no correlation was found between air pollution and noise (Figure S4). The absence of correlation can be explained by the dynamic nature of the measurements, because noise and air pollution may have different spatial patterns.

Single- and multi-exposure models

The air pollution components PM_2.5 and NO₂ were positively associated with dyslipidemia (RR = 1.12 [95% CI, 1.02-1.22]; RR = 1.15 [95% CI, 1.07-1.23], respectively), obesity (RR = 1.15 [95% CI, 1.02-1.30]; RR = 1.19 [95% CI, 1.05-1.35], respectively), and MetS (RR = 1.16[ 95% CI, 0.96-1.41]; RR = 1.25 [95% CI, 1.04-1.50], respectively), even after adjustment for individual and other environmental characteristics (ie, greenness, neighborhood economic level, and noise) (Table 2). Additionally, aircraft noise exposure was associated with increased MetS risk (RR = 1.18; 95% CI, 1.05-1.31). No association was observed between road noise exposure and the 4 cardiometabolic outcomes.

Table 2.

Association between dynamic pollution exposures (air and noise pollution) and each cardiometabolic disorder^a

Pollution exposure	Dyslipidemia b	Hypertension b	Obesity b	Metabolic syndrome b
Air pollution
PM2.5	1.12 (1.02-1.22)	1.00 (0.89-1.13)	1.15 (1.02-1.30)	1.16 (0.96-1.41)
NO2	1.15 (1.07-1.23)	1.02 (0.89-1.19)	1.19 (1.05-1.35)	1.25 (1.04-1.50)
Transportation noise
Aircraft	1.03 (0.95-1.11)	1.02 (0.89-1.17)	1.05 (0.94-1.18)	1.18 (1.05-1.31)
Road	1.00 (0.93-1.06)	1.01 (0.85-1.13)	0.96 (0.86-1.07)	0.95 (0.84-1.08)

^aResults from fully adjusted models are presented (model 4). Models were adjusted for age, sex, ethnicity, income, greenness, neighborhood economic level, and secondary exposure (air pollution or noise exposure).

^bData are reported as relative risk (95% CI).

Abbreviations: NO₂, nitrogen dioxide; PM_2.5, fine particulate matter.

When the models were fully adjusted (model 4), the results remained consistent with the other models (Figure 1). We have established the suitability of a linear relationship for this analysis by examining various spline models (Table S2). No association was found between static exposures and cardiometabolic disorders (Figure S5).

Figure 1.

Association between dynamic exposures and each cardiometabolic disorder, for dynamic exposure to fine particulate matter (PM_2.5) (A), nitrogen dioxide (NO₂) (B), aircraft noise (C), and road noise (D). Model 1 was a crude model; model 2 was adjusted for age, sex, ethnicity, and income; model 3 was further adjusted for greenness and neighborhood economic level; model 4 was additionally adjusted for secondary exposure (air pollution or noise exposure). MetS, metabolic syndrome.

Interaction

Two multiplicative interactions were statistically significant (P < .10): between NO₂ and aircraft noise regarding obesity (P = .04), and between NO₂ and road noise regarding MetS (P = .08). All the other interaction terms were nonsignificant. Overall, in stratified analysis, the RR was lower among participants less exposed to air pollution components (ie, exposed to PM_2.5 or NO₂ levels below the median), and higher among participants exposed to higher air pollution levels (Figure 2). Specifically, we found that aircraft noise was associated with an increased risk of MetS and obesity among participants exposed to high PM_2.5 levels (Figure 2A) and participants exposed to high NO₂ levels (Figure 2B). Finally, road noise was associated with a lower risk of obesity among participants exposed to low NO₂ levels (Figure 2B). The Cochran Q test showed that the association between aircraft noise and obesity was significantly different according to NO₂ level (Cochran Q P = .09). Moreover, the association between aircraft noise and MetS was significantly different according to PM_2.5 level (Cochran Q P = .11).

Figure 2.

Association between dynamic noise exposures and cardiometabolic disorders according to fine particulate matter (PM_2.5) (A) or nitrogen dioxide (NO₂) (B) exposure level (according to the median). Models were adjusted for age, sex, ethnicity, income, greenness, and neighborhood economic level. MetS, metabolic syndrome.

Lastly, the association between road noise and obesity and hypertension varied according to NO₂ level (Cochran Q P = .01 and .11, respectively). For all 4 combinations of air pollution components and transportation noise (ie, aircraft noise × PM_2.5; aircraft noise × NO₂; road noise × PM_2.5; road noise × NO₂), 2 RERI values were statistically significant. The RERIs for aircraft noise and NO₂ on obesity and MetS (0.13 [95% CI, 0.03-0.22] and 0.13 [95% CI, 0.02-0.25], respectively) were higher than 0, indicating the presence of an additive interaction (Figure 3).

Figure 3.

Interaction between dynamic exposures on the additive scale (using relative excess risk due to interaction [RERI]). MetS, metabolic syndrome; NO₂, nitrogen dioxide; PM_2.5, fine particulate matter.

Discussion

In an ethnically diverse population living in San Diego County, CA, we found that dynamic exposure (using GPS data) to PM_2.5 was associated with an increased risk of dyslipidemia and obesity, and dynamic exposure to NO₂ was associated with an increased risk of dyslipidemia, obesity, and MetS. Although road traffic noise was not associated with any of the cardiometabolic disorders, aircraft noise was associated with an increased risk of MetS. The associations between air pollution and metabolic disorders were independent of transportation noise in our study, and vice versa. On the multiplicative scale, we found statistically significant interactions between air pollution and noise. Exposure to aircraft noise was associated with an increased risk of obesity and metabolic disorders only among participants exposed to high levels of air pollution. The findings obtained on the additive scale were in line with those obtained on the multiplicative scale: we found that people exposed to high levels of air pollution seem to be more affected by aircraft noise for obesity and MetS. Our findings suggested that air pollution and aircraft noise may have synergistic effects on obesity and MetS.

Both PM_2.5 and NO₂ were statistically significantly associated with dyslipidemia, before and after adjustment for transportation noise exposure. These findings are consistent with the results highlighted in a 2020 meta-analysis.¹³ However, all 4 studies mentioned in that meta-analysis only looked at some of the conditions that constitute dyslipidemia, and none of them adjusted for transportation noise. The present results also suggest that greater exposure to PM_2.5 and NO₂ is associated with a higher risk of obesity and MetS, which is in line with findings from previous studies,^55-58 even if the evidence is not clear for MetS.⁵⁹

We also found that exposure to aircraft noise was associated with increased MetS risk. Previous studies that analyzed the effect of noise on the incidence of MetS focused either on road traffic noise or participants’ noise perception.^36,60 Our results on hypertension are partially aligned with those of the recent review of Münzel et al.¹⁴ indicating weak or no association with hypertension. However, in our study, no association was found between road noise and any of the outcomes. In California, Yu et al.³⁶ found elevated noise levels were positively associated with new occurrences of MetS. The main difference between the Yu et al.³⁶ study and ours is that their cohort included Mexican Americans aged 60 to 101 years. The lack of association between road noise and cardiometabolic outcomes can potentially be explained by the continuous nature of road noise (as opposed to the intermittent nature of aircraft traffic noise) and timing. Additionally, aircraft noise can be more disruptive than road noise. Indeed, studies have demonstrated that even at the same level of day-evening-night noise, individuals tend to be more annoyed by aircraft noise than road traffic noise.⁶¹ Additionally, noise barriers, such as high-quality windows, can help block road noise exposure, but not aircraft noise.⁶² Furthermore, a 2022 meta-analysis found that populations exposed to high levels of aircraft noise may be at greater risk of sleep disturbance, which is a potential major mechanism between noise and cardiometabolic diseases.⁶²

This study represents 1 of the first attempts to comprehensively investigate the interaction between air pollution and transportation noise exposures, at both multiplicative and additive scales, to better understand their combined impact on 4 different cardiometabolic disorders (dyslipidemia, hypertension, obesity, and MetS). To our knowledge, only 2 other studies have looked at multiplicative interaction between these variables; 1 study showed a larger effect,³⁵ and the other did not find a statistically significant interaction.³⁴ Our results suggest that dynamic exposure to PM_2.5 and NO₂ modifies the association between aircraft noise and MetS. Specifically, we found that the effect of aircraft noise on the risk of MetS seems higher for participants with greater exposure to air pollution. Additionally, NO₂ modified the association between aircraft noise and obesity. The mechanisms explaining these results may be partly given by the fact that the noise emitted by aircrafts is louder than road traffic and less continuous, which could enhance stress in individuals⁶¹ already exposed to higher oxidative stress by air pollution. Another possible mechanism is the effect of noise on sleep disturbance and stress. As mentioned, individuals exposed to high levels of aircraft noise may be at greater risk of sleep disturbance, which may cause metabolic changes. Combined with the oxidative stress and low-grade local and systemic inflammation resulting from air pollution, this may lead to an increased risk of MetS and obesity.⁶³

Strengths and limitations

Strengths of this study include the use of dynamic exposure data (based on GPS data), which more accurately reflect individual-level exposure. Notably, our findings revealed that although there were no significant associations between static exposure data and the outcomes of interest, the use of dynamic exposure data highlighted meaningful relationships. This is an important finding because it underscores the need for more accurate exposure assessments in epidemiologic research. The study also relied on objective and standardized anthropometric measures, as well as biomarkers from blood samples, increasing the reliability of our results. Furthermore, our sample population was diverse in terms of ethnicity, including participants underrepresented in epidemiologic research, adding to the generalizability of our findings. Finally, exposure misclassification was minimized in our study, because participants lived at their same address for an average of 17 years.

However, there are some limitations to acknowledge. First, the sample size is relatively small, limited by the collection of cardiometabolic biomarkers and GPS and accelerometer data. Second, noise, air pollution, and health indicators were assessed at a single point in time, which may not fully capture the complex relationship between these factors throughout the course of one’s life. Additionally, other sources of environmental noise were not considered, such as noise from railways or industrial sites, co-exposure to occupational noise, or military activity (which is particularly important in parts of San Diego County). We also did not consider the potential modification of noise exposure by factors such as building structures and materials, closing of windows, indoor noise level, or having the bedroom facing the backyard, which were not available in our data. Finally, we used the 24-hour equivalent sound level noise metric to evaluate noise effects, which is an average energetic dose over longer periods and did not incorporate a nighttime noise penalty to account for nocturnal environmental noise. This is an important limitation because nighttime noise can have a greater impact on health through disruptions to sleep.⁶⁴ In addition, a more precise assessment of road noise exposure (ie, using more sophisticated methods, considering long-term measurements and numerous environmental factors) should be used in future studies, as done recently.⁶⁵

Conclusion

Air pollution and aircraft noise interact to affect risk of obesity and MetS at multiplicative and additive scales, suggesting a potential synergism between these 2 exposures. These findings highlight the importance of minimizing exposure to both aircraft noise and air pollution. Future studies should evaluate air and noise pollution as effect modifiers when studying their impacts on health.

Acknowledgments

We thank Maren Hale, University of California, San Diego, for her helpful feedback on this manuscript. We thank the US Department of Transportation Bureau of Transportation Statistics, which provided the road and aircraft noise layers for California.

Supplementary material

Supplementary material is available at the American Journal of Epidemiology online.

Funding

This work was supported by the National Cancer Institute (grants R01CA228147, R01CA179977).

Conflict of interest

The authors declare no conflicts on interest.

Data availability

Data will be made available upon reasonable request.