Traffic-Related Pollutants: Exposure and Health Effects Among Hispanic Children

Abstract

We examined traffic-related pollution (TRP) exposure and respiratory health effects in Hispanic white (HW) children, both compared with non-Hispanic white (NHW) children and according to genetically determined Native American (NA) ancestry. The sample included over 5,000 children from the Children’s Health Study in California, followed during 1993–2014. HW children were 1.47 (95% confidence interval (CI): 1.24, 1.73) times more likely to live close (<500 m) to a freeway and 1.54 (95% CI: 1.26, 1.87) times more likely to live close (<75 m) to a major nonfreeway road compared with NHW children. Among HW children, those with >50% NA ancestry were >40% more likely to live close to a freeway or to a major nonfreeway road, compared with those with ≤50% NA ancestry. The association of TRP with ever having been diagnosed by a doctor as having asthma differed between HW and NHW children (P < 0.05), with the strongest association among HW children with >50% NA ancestry. Within this subgroup, those close to a major nonfreeway road were 2.16 (95% CI: 1.26, 3.69) times more likely to have ever reported asthma compared with those living further away. This paper provides evidence that HW children in southern California, especially those with greater NA ancestry, are more exposed to TRP and are potentially at greater risk for TRP-related respiratory health effects.

In recent years, significant changes have occurred in the regulation of air pollution in the United States. A primary reason for more stringent regulatory standards is the accrued evidence from epidemiologic studies on the negative impact of air pollution with respect to respiratory health (1–8). A recent study suggested that even with improved regulations to reduce pollution levels, multiple areas in the United States still suffer from poor air quality (6). A number of these high-pollution sites are found in Los Angeles County and surrounding communities. Previous studies have also demonstrated an increased burden of air pollution on children and adolescents in metropolitan areas (1, 3, 4, 8, 9).

The University of Southern California Children’s Health Study (CHS) was initiated in 1993 with the overall goal of investigating the effects on children’s respiratory health of air pollution in southern California (10, 11). Over the course of this study, various pollutants were found to be associated with reduced lung-function development and increased incidence of adverse respiratory outcomes including asthma (2, 5, 9, 12, 13). Pollutants linked to these adverse events include particulate matter (particulate matter having an aerodynamic diameter of ≤10 μm or ≤2.5 μm), nitrogen dioxide (NO₂), and ozone (O₃). The concentrations of these pollutants vary on a local and regional basis. Fixed-site monitors have characterized long-term regional levels of these pollutants across southern California, while local variation has been determined primarily by indirect measurements focused on indicators of traffic-related pollution (TRP), including distance to major roadways and traffic volume on nearby roads.

Racial/ethnic disparities in air-pollution exposure and health outcomes have been noted in a number of studies conducted throughout the United States (1, 3, 6, 14–20). In southern California, Hispanic people make up a large proportion of the population in many communities. A study conducted in Long Beach found that Hispanic residents were exposed to higher vehicle traffic exposure, as measured by vehicle miles traveled, than were non-Hispanic white (NHW) residents (16). Another study within Los Angeles determined that higher NO₂ concentrations were significantly associated with neighborhoods and parks that have larger proportions of Hispanic residents (19). Hispanic people in southern California are an admixed population, ranging from individuals with nearly 100% Native American (NA) ancestry to individuals with nearly 100% European ancestry—most with significant contributions from both of these ancestral populations. None of the studies listed above has investigated both exposure and health effects in Hispanics, or in particular how intra-Hispanic heterogeneity relates to pollution exposure or how it might modify the relationship between pollution and health effects.

In this study, we investigated whether disparities in air-pollution exposure by ethnicity and ancestry exist within 14 southern California communities. Specifically, we examined whether Hispanic white (HW) children were more exposed to local TRP compared with NHW children. We also examined whether variation in air-pollution exposure existed based on estimated proportion of NA ancestry among HW children, and we explored factors that might partially explain disparities in exposure. Finally, we analyzed the association between TRP exposure and asthma, specifically how this association varied depending on ethnicity and ancestry.

METHODS

Study participants

The CHS is a prospective longitudinal study that recruited over 12,000 children across 16 communities during 1993–2003. The children were recruited from public elementary schools in their respective communities and were followed for up to 13 years, with the last cohort followed until 2014. Upon study entry, children received a baseline questionnaire and informed consent form, completed by their parent or legal guardian. Annual follow-up questionnaires were also completed by the child’s parent or legal guardian to monitor changes in time-varying exposures and outcomes. In this report, we focused on near-roadway exposure and health effects. Because all participants in 2 communities, Lake Gregory (n = 914) and Lompoc (n = 460), lived at least 10 km from any major roadways, they were excluded from this analysis. Among the remaining communities, 2,713 NHW and 2,892 HW participants had genetic data collected to determine ancestral origin. Of the 5,605 participants with genetic data, 268 did not have valid geocode data available for their residence, leaving 2,570 NHW and 2,767 HW participants for analysis.

The baseline questionnaire, completed at study entry by each child’s parent or legal guardian, was used to obtain data on race, Hispanic ethnic origin, total household income, parental education, history of parental asthma, and smoking in the household. “Hispanic white” was defined as selecting “Yes” for Hispanic origin and not selecting “Asian” or “African American” for race. “Non-Hispanic white” was assigned to any participant who answered “White” for race and “No” for Hispanic origin. Additional details of the CHS study subjects have been published previously (5, 10).

Computation of ancestry coefficients

In a previous candidate-gene study, ancestral origins were determined for CHS participants with available DNA. Through the use of 233 ancestral informative markers and the program STRUCTURE (21), the percentage of European and NA ancestry was estimated for each participant in the subset. For analysis, the proportion of NA ancestry was categorized into 2 groups for all study participants: less than or equal to 50% (low-NA) and greater than 50% (high-NA) NA ancestry.

Determination of traffic exposure data

We characterized exposure of every study participant to traffic-related pollutants at their baseline residence by 4 measures: proximity of the child’s residence to the nearest freeway, proximity to the nearest major nonfreeway road, estimated freeway-related nitrogen oxide (NO_x) exposure, and estimated nonfreeway-related NO_x exposure.

Participant baseline addresses were standardized and geocoded using Tele Atlas MultiNet road network data (Tele Atlas Inc., Menlo Park, California). ArcGIS software (ESRI, Redlands, California) was used to estimate distance to the nearest freeway and to the nearest major nonfreeway road. Residential NO_x exposure to local freeway and nonfreeway major roads, accounting for traffic volume, wind speed, wind direction, and other factors, was estimated using the CALINE4 line source dispersion model (22). Additional details of traffic-related exposure assessment can be found in the online supplement (Web Appendix 1, available at https://academic.oup.com/aje).

Assessment of asthma

Three different asthma outcomes were considered in the analysis, including baseline-prevalent asthma, incident asthma, and ever asthma. The parent or legal guardian of the child reported on the baseline questionnaire whether or not the child had ever been diagnosed by a doctor as having asthma and, on the annual follow-up questionnaires, whether their doctor-diagnosis of asthma status changed during the study. “Baseline-prevalent” asthma was defined as a “Yes” answer to the question on the baseline questionnaire. “Incident” asthma was defined as a “No” answer on the baseline questionnaire and a “Yes” answer on a subsequent follow-up questionnaire. Any cases of incident asthma that occurred after a move from the child’s baseline residence were excluded from the analysis. Finally, a child was determined to have “ever asthma” if they had either baseline-prevalent or incident asthma.

Statistical analysis

For analysis of pollution exposure, the outcome data consisted of 4 measures of residential TRP exposure determined at study entry. The outcome measures were: distance to nearest major freeway, distance to nearest major nonfreeway road, predicted exposure levels to NO_x due to freeways, and predicted exposure levels to NO_x due to major nonfreeway roads.

Multiple linear regression was used to assess whether HW children were more exposed to TRP than NHW and to examine how exposure varied by categories of NA ancestry. To meet regression assumptions, a cube root transformation was used on distance to freeway and major nonfreeway roads, and a log transformation was used for both freeway and major nonfreeway NO_x exposures. A Box-Cox transformation was used to confirm the appropriateness of these transformations. All models included adjustment for study community and cohort as fixed effects. Previous results demonstrated an adverse association between living within 500 m of a major freeway (2) or 75 m of a major nonfreeway road (5) and asthma. We used logistic regression to determine the odds ratio for living within these 2 road buffers based on Hispanic ethnicity and on NA ancestral group.

Logistic regression was then used to model the relationship between asthma status and exposure to TRP, within subgroups defined by ethnicity (HW vs. NHW) and ancestry (within HW: high-NA vs. low-NA). We focused these health analyses on children who lived their entire lives at the same home (“lifelong” residents). All models included adjustment for age at baseline recruitment, community, and sex. We added an interaction term to the logistic model to determine whether the association between TRP and asthma differed by ethnicity in all participants and by NA ancestry within HW participants. Sensitivity analyses were completed within each stratum to assess potential confounders of the TRP-asthma associations. Confounding was evaluated by determining whether the regression coefficient changed by >15% after addition to the basic model. The potential confounders we evaluated included socioeconomic status based on family household income and highest parental education level, self-reported doctor-diagnosed asthma in either parent, and whether or not smoking occurred in the participant’s household.

All analyses were completed using SAS, version 9.4 (SAS Institute, Inc., Cary, North Carolina). Tests of statistical significance were based on a 0.05 significance level, assuming a 2-sided alternative hypothesis.

RESULTS

Traffic proximity

There were nearly equal proportions of HW and NHW participants in the overall sample (Table 1). The vast majority of NHW subjects (99.5%) fell in the low-NA group. In contrast, HW subjects were more heterogeneous, with 68% in the high-NA group and 32% in the low-NA group. At study entry, doctor-diagnosed asthma was reported for 13.2% of HW children, significantly less than for NHW children (16.8%). Incident asthma during the study was lower but not significantly different in HW children (11.7%) compared with NHW children (12.4%). HW participants tended to be from families with lower total household income and parental education (see Web Table 1).

Table 1.

Relationship of Hispanic Descent to Ancestry, Asthma, and Lifelong Residency Among Participants in the Children’s Health Study, California, 1993–2014

Variable	Ethnicity				P Value
	Non-Hispanic White n = 2,570 (48.3%)		Hispanic White n = 2,767 (51.7%)
	No. of Participants	%	No. of Participants	%
Native American ancestry					<0.0001
≤50%	2,556	99.5	874	31.6
>50%	14	0.5	1,893	68.4
Prevalent asthma					0.0002
Yes	432	16.8	365	13.2
No	2,137	83.2	2,398	86.7
Missing	1	0.04	4	0.14
Incident asthmaa					0.45
Yes	266	12.4	281	11.7
No	1,871	87.5	2,117	88.1
Missing	1	0.05	4	0.17
Ever had asthma					0.0014
Yes	698	27.2	646	23.4
No	1,871	72.8	2,117	76.5
Missing	1	0.04	4	0.14
Lifelong resident					0.022
Yes	742	28.9	791	28.6
No	1,215	47.3	1,499	54.2
Missing	613	23.8	477	17.2

^a Excludes children who had prevalent asthma at study entry.

On average, NHW participants lived 1,480 m from a freeway, while HW participants lived 184 m (95% confidence interval (CI): 117, 252) closer (Figure 1A). Among HW participants, high-NA children lived 165 m (95% CI: 81, 249) closer to a freeway on average than did low-NA children. Analogous to freeway distance, the mean distance to major nonfreeway roads among NHW participants was 395 m (Figure 1B), 71 m (95% CI: 48, 94) farther away than the mean distance among HW participants (Figure 1B). Within the HW group, high-NA participants lived 48 m (95% CI: 21, 76) closer to a major nonfreeway road than did low-NA participants.

Figure 1.

Association of traffic-related pollution exposure with Hispanic ethnicity and percentage of Native American ancestry among participants in the Children’s Health Study, California, 1993–2014. The graphs show mean estimates for distance to nearest freeway (A), distance to nearest major nonfreeway road (B), estimated freeway nitrogen oxide (NO_x) exposure in parts per billion (ppb) (C), and estimated nonfreeway NO_x exposure in ppb (D), according to self-identified Hispanic descent in all participants and, among Hispanic white participants only, percentage of Native American ancestry. Data are based on the baseline residence for each participant. The model adjusted for community and study cohort. Error bars are 95% confidence intervals for mean estimates.

Similar trends were observed for predicted exposure levels to freeway NO_x (parts per billion) and nonfreeway road NO_x exposure (Figure 1C and 1D). On average, predicted exposure levels were 14% (95% CI: 8, 20) higher for freeway NO_x and 18% (95% CI: 14, 22) higher for nonfreeway NO_x in HW children compared with NHW children. By ancestry status, high-NA HW children were found to have 16% (95% CI: 9, 24) higher predicted freeway NO_x exposure and 16% (95% CI: 11, 22) higher predicted nonfreeway road NO_x levels than were low-NA HW children.

HW participants were also more likely to live within critical buffers around freeways and major nonfreeway roads (Table 2). HW children were 1.47 times more likely to live within 500 m of a freeway (95% CI: 1.24, 1.73) than were NHW children, and also 1.54 times more likely to live within 75 m of a major nonfreeway road (95% CI: 1.26, 1.87). Among HW participants, high-NA children were 1.41 times more likely to live within 500 m of a freeway (95% CI: 1.12, 1.78) and 1.45 times more likely to live within 75 m of a freeway (95% CI: 1.10, 1.90) than were low-NA children.

Table 2.

Association of Ethnicity and Percentage of Native American Ancestry With Living Within 500 m of a Freeway and Residing Within 75 m of a Major Nonfreeway Road Among Participants in the Children’s Health Study, California, 1993–2014

Variable	No. of Participantsa	Residence Within 500 m of Freeway		Residence Within 75 m of Major Nonfreeway Road
		ORb	95% CI	ORb	95% CI
Ethnicity
NHW	2,570	1.00	Referent	1.00	Referent
HW	2,767	1.47c	1.24, 1.73	1.54c	1.26, 1.87
NA ancestry (HW only)
≤50%	874	1.00	Referent	1.00	Referent
>50%	1,893	1.41d	1.12, 1.78	1.45d	1.10, 1.90

Abbreviations: CI, confidence interval; HW, Hispanic white; NA, Native American; NHW, non-Hispanic white; OR, odds ratio.

^a Total number of participants in subgroup.

^b ORs adjusted for community and cohort.

^cP ≤ 0.001.

^dP ≤ 0.01.

HW participants tended to be from families with lower total household income and parental education. Adjustment for either parental education or total household income attenuated the association of Hispanic descent and NA ancestry among HW participants with all four TRP indicators (see Web Tables 2 and 3). In general, income had a larger effect than education on the estimates and overall significance, with many regression estimates halved in magnitude compared with the unadjusted base model. Addition of both parental education and total household income to models using reported Hispanic descent reduced all regression estimates, except distance to nearest major roadway, to nonsignificant levels (see Web Table 2). However, all estimates still maintained the same direction of association, with HW and >50%-NA subjects living closer to major roadways and having higher predicted NO_x exposure.

Asthma associations

Thus far, we have reported that, HW children, especially those with higher estimated NA ancestry, were more likely to live closer to major roads than NHW children. As shown in Table 3, we also found that among lifelong residents, the association between traffic proximity and asthma risk was highest among HW children, and in particular high-NA HW children. By ethnicity, compared with living further than 75 m from a major nonfreeway road, HW children within 75 m of a major nonfreeway road were 2.10 (95% CI: 1.30, 3.39) and 2.20 (95% CI: 1.14, 4.25) times more likely to have reported ever having asthma and incident asthma, respectively. In contrast, NHW children within 75 m of major nonfreeway road were 0.91 (95% CI: 0.51, 1.65) and 0.81 (95% CI: 0.33, 1.99) as likely to have reported ever having asthma and incident asthma, respectively. Stratification by NA ancestry within HW participants revealed that the most significant associations were in the high-NA group. High-NA HW children living within 75 m of a major nonfreeway road were 2.16 (95% CI: 1.26, 3.69) and 2.72 (95% CI: 1.30, 5.73) times more likely to have reported ever having asthma and incident asthma, respectively. Additionally, HW children living within 75 m of a major nonfreeway road were 1.77 (95% CI: 1.00, 3.15) times more likely to report prevalent asthma at study entry compared with those at least 75 m away, while the corresponding odds ratio for NHW children was 0.96 (95% CI: 0.48, 1.90). Interestingly, the odds ratio for prevalent asthma in low-NA HW children, 2.49 (95% CI: 0.75, 8.29), was higher than in high-NA children, 1.72 (95% CI: 0.90, 3.30), although the difference was not statistically significant. Similar associations of asthma with estimated nonfreeway NO_x exposure, stratified by either Hispanic descent or NA ancestry within HW children, were observed (see Web Table 4). The associations of the asthma outcomes with distance to a major nonfreeway road and with nonfreeway NO_x exposure were very similar after adjustment for total household income, parental education, history of parental asthma, and smoking in the child’s household (Table 4 and Web Table 5).

Table 3.

Association of Living Within 75 m of a Major Nonfreeway Road With Asthma, Stratified by Ethnicity and by Estimated Percentage of Native American Ancestry Among Lifelong Residents Participating in the Children’s Health Study, California, 1993–2014

Variable	No. of Participantsa	Ever Asthma		Prevalent Asthma		Incident Asthma
		ORb	95% CI	ORb	95% CI	ORb	95% CI
Ethnicity
NHW	741	0.91	0.51, 1.65	0.96	0.48, 1.90	0.81	0.33, 1.99
HW	790	2.10c	1.30, 3.39	1.77	1.00, 3.15	2.20c	1.14, 4.25
P valued		0.03e		0.18		0.077
NA ancestry (HW only)
≤50%	242	2.04	0.69, 6.08	2.49	0.75, 8.29	1.31	0.24, 7.01
>50%	548	2.16c	1.26, 3.69	1.72	0.90, 3.30	2.72c	1.30, 5.73
P valuef		0.93		0.60		0.43

Abbreviations: CI, confidence interval; HW, Hispanic white; NA, Native American; NHW, non-Hispanic white; OR, odds ratio.

^a Total number of participants in subgroup with asthma data available.

^b Odds ratios for the corresponding asthma outcome versus living within 75 m of a major road, adjusted for age at baseline, study community, and sex.

^cP ≤ 0.01.

^d Test for multiplicative interaction between living within 75 m of a major roadway and ethnicity.

^eP ≤ 0.05.

^f Test for multiplicative interaction between living within 75 m of a major roadway and estimated percentage of NA ancestry.

Table 4.

Association of Living Within 75 m of a Major Nonfreeway Road With Asthma, Stratified by Ethnicity and by Estimated Percentage of Native American Ancestry, Adjusted for Potential Confounding Factors, Among Lifelong Residents Participating in the Children’s Health Study, California, 1993–2014

Outcome	Base Modela		Base + Parental Education		Base + Household Income		Base + Parental Asthma		Base + Household Smoking
	ORb	95% CI	ORb	95% CI	ORb	95% CI	ORb	95% CI	ORb	95% CI
Ever asthma
NHW	0.91	0.51, 1.65	0.92	0.51, 1.65	0.99	0.53, 1.86	0.98	0.54, 1.78	0.86	0.47, 1.56
HW	2.10	1.30, 3.39	2.08	1.26, 3.42	1.90	1.12, 3.24	2.14	1.27, 3.60	1.98	1.21, 3.22
≤50% NA	2.04	0.69, 6.08	2.10	0.70, 6.29	1.51	0.48, 4.73	1.56	0.43, 5.58	2.05	0.70, 6.04
>50% NA	2.16	1.26, 3.69	2.04	1.17, 3.54	2.05	1.13, 3.71	2.27	1.29, 4.01	1.98	1.15, 3.42
Prevalent asthma
NHW	0.96	0.48, 1.90	0.99	0.50, 1.96	0.97	0.47, 2.02	1.00	0.50, 2.02	0.86	0.42, 1.75
HW	1.77	1.00, 3.15	1.72	0.95, 3.10	1.65	0.86, 3.14	1.87	1.00, 3.49	1.72	0.96, 3.11
≤50% NA	2.49	0.75, 8.29	2.42	0.72, 8.16	2.32	0.68, 7.95	2.17	0.53, 8.93	2.48	0.74, 8.29
>50% NA	1.72	0.90, 3.30	1.59	0.82, 3.09	1.47	0.70, 3.09	1.90	0.96, 3.74	1.65	0.85, 3.21
Incident asthma
NHW	0.81	0.33, 1.99	0.80	0.33, 1.97	0.86	0.35, 2.15	0.88	0.36, 2.18	0.82	0.33, 2.01
HW	2.20	1.14, 4.25	2.24	1.12, 4.46	2.11	1.03, 4.30	2.17	1.05, 4.50	2.05	1.05, 4.02
≤50% NA	1.31	0.24, 7.01	1.81	0.37, 8.95	0.71	0.11, 4.63	1.04	0.15, 7.12	1.41	0.29, 6.78
>50% NA	2.72	1.30, 5.73	2.65	1.26, 5.60	2.70	1.25, 5.80	2.76	1.28, 5.97	2.47	1.19, 5.14

Abbreviations: CI, confidence interval; HW, Hispanic white; NA, Native American; NHW, non-Hispanic white; OR, odds ratio.

^a Base model adjusted for age at baseline, study community, and sex.

^b Odds ratios for the corresponding asthma outcome versus living within 75 m of a major road.

The general trends observed for distance to a major nonfreeway road were also observed for distance to a freeway, although the effect magnitudes were lower (see Web Table 6). For example, among high-NA HW children, those within 500 m of a freeway were 1.20 (95% CI: 0.74, 1.94) and 1.42 (95% CI: 0.71, 2.83) times more likely to have reported ever having asthma or incident asthma, respectively, compared with those living at least 500 m from a freeway. These findings were also very similar after adjustment for total household income, parental education, history of parental asthma, and smoking in the child’s household (data not shown).

DISCUSSION

We have demonstrated that in southern California, HW children lived closer to freeways and major nonfreeway roads and had higher predicted exposures to traffic-related NO_x than NHW children. Our use of ancestral data allowed for further refinement of TRP exposure differences within the HW subpopulation. For example, traffic exposure among HW children with >50% NA ancestry (high-NA) was significantly higher than among HW children with ≤50% NA ancestry (low-NA). Additionally, high-NA HW children were most likely to live within a critical distance, with regard to respiratory health, of freeways and major nonfreeway roads.

McConnell et al. (5) reported evidence that residential proximity to a major roadway (<75 m) was associated with increased risk of asthma among lifelong residents. We further evaluated this association and determined that the association of traffic with asthma differed by reported Hispanic descent. Interestingly, overall asthma prevalence and incidence were lowest among HW children and those with higher NA ancestry. However, the associations of living within 75 m of a major nonfreeway road with ever having asthma and incident asthma were significant only among these same subgroups. This finding may highlight a difference in the importance of genetic and environmental cofactors by ancestry. Given the higher asthma rates among lower-NA children, asthma susceptibility in these subgroups might be high regardless of TRP exposure levels. In contrast, high-NA children might have a lower genetic susceptibility, thus making the association of environmental factors, such as TRP, more evident when exposure disparities do exist. We did not find significant associations between living within 500 m of a freeway and any of the asthma outcomes (data not shown), a finding consistent with the analysis of the overall sample in McConnell et al. (5).

Biologically, a potential causal relationship of TRP with asthma is supported by previous studies that have investigated the role of air pollutants as a source of oxidative stress in the pathogenesis of asthma (23, 24). In bronchial asthma, oxidative stress promotes inflammatory mediators due to increased airway aggravation, which in turn can lead to increased hypersensitivity. Further research is needed to determine whether the biological response to increased oxidative stress differs by ethnicity or ancestry due to differences in cofactors that could include genetic susceptibility, other environmental exposures, and social factors.

Our finding of higher TRP exposure among HW children compared with NHW children aligns with previous studies completed in the Los Angeles area and across the United States (14–16, 18, 19). Several of these studies made similar adjustments for socioeconomic indicators and found these factors to provide a potential explanation for exposure disparities. Although income and education partly explained differences in traffic exposure by ethnicity and ancestry, these factors had little impact on estimates of association between traffic and asthma within HW and NA subgroups. Additional sensitivity analyses of potential confounders that include history of parental asthma and second-hand smoke exposure did not significantly change our results either. Previous studies have shown differences by ethnicity in the relationship of TRP exposure and hospitalization due to asthma (3, 15, 20). However, those associations were partially explainable by lower insurance enrollment, decreased access to care, and other socioeconomic factors that were correlated with TRP exposure. Thus, although socioeconomic factors might confound the association between TRP exposure and asthma management, they do not appear to have the same effect on how traffic exposure influences asthma risk. Despite completing analysis only on those CHS subjects with ancestral data, the size of the sample and diversity of the study communities leads to increased generalizability of the results.

There are limitations in the interpretation of our results and potential bias with regard to the study outcomes. The assessment of doctor-diagnosed asthma by questionnaire is subject to recall bias, more so for prevalent asthma reported at baseline than incident asthma during study follow-up. However, previous work has shown agreement between self-reported doctor-diagnosed asthma and true diagnosis in adults (25). The reliability of parent-reported asthma might have differed by ethnicity as well. However, the reliability would need to be differential with respect to both ethnicity and TRP exposure to affect our analyses, and this is unlikely. Diagnostic bias could also lead to a higher likelihood of asthma diagnosis in communities known for high air pollution or in children with known environmental exposures (i.e., parental smoking), which we attempted to control for in our sensitivity analysis.

Another consideration is the association of parental asthma with Hispanic descent. NHW children were more likely to have at least 1 parent with a self-reported history of asthma. Parents with a history of asthma might choose to live farther away from major roadways for their health and their children’s health. Potential self-selection by NHW parents to live further from roadways might bias the association of TRP with asthma towards the null within this subgroup. However, we did not find a significant association between parental asthma and living within 75 m of a major nonfreeway road, suggesting that this bias did not affect our observed associations between TRP exposure and childhood asthma.

Last, our results might be limited by the accuracy with which the true unobserved exposure to traffic-related pollution was captured by distance-to-road measures. Specifically, for 2 children living the same distance from a road, the true TRP exposure is likely to be different due to differences in meteorology, types of vehicles, and other factors. The general agreement between our results based on distance measures and the CALINE4 model-based estimates, which incorporate several of these additional factors, provides evidence that we have captured reasonable estimates of TRP exposure.

Our results demonstrate the importance of understanding how air-pollution exposure varies with ethnicity as determined by both self-identification and by genetically determined ancestry. Additional work is needed to understand other factors that affect a child’s exposure to air pollution, such as time-activity patterns, and how those might also vary by ethnicity or genetic ancestry. An increased understanding of exposure differences across and within racial/ethnic subgroups will be needed to recognize those populations at greatest risk of health effects due to air pollution.

Abbreviations

CHS

Children’s Health Study

CI

confidence interval

high NA

estimated percentage of Native American ancestry is greater than 50%

HW

Hispanic white

low NA

estimated percentage of Native American ancestry is less than or equal to 50%

NA

Native American

NHW

non-Hispanic white

NO_x

nitrogen oxide

TRP

traffic-related pollution