UBC Faculty Research and Publications

Chronic Traffic-Related Air Pollution and Stress Interact to Predict Biologic and Clinical Outcomes in… Chen, Edith; Schreier, Hannah M. C.; Strunk, Robert C.; Brauer, Michael Jul 31, 2008

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
52383-Chen_E_et_al_Chronic_traffic_air_pollution.pdf [ 133.32kB ]
Metadata
JSON: 52383-1.0220725.json
JSON-LD: 52383-1.0220725-ld.json
RDF/XML (Pretty): 52383-1.0220725-rdf.xml
RDF/JSON: 52383-1.0220725-rdf.json
Turtle: 52383-1.0220725-turtle.txt
N-Triples: 52383-1.0220725-rdf-ntriples.txt
Original Record: 52383-1.0220725-source.json
Full Text
52383-1.0220725-fulltext.txt
Citation
52383-1.0220725.ris

Full Text

970 VOLUME 116 | NUMBER 7 | July 2008 • Environmental Health PerspectivesResearch | Children’s HealthBoth the physical and the social environ-ments have long been hypothesized to beimportant contributors to childhood asthma.Although research identifying the physicaland social environmental factors that play arole in asthma has made great strides (Crainet al. 2002; Wright 2005), much of thisresearch exists independently of one another.In contrast, few studies have examined howthe physical and social environments mightinteract to affect asthma outcomes. For exam-ple, certain combinations of physical andsocial environmental exposures might pro-duce unique effects beyond exposure to oneof these factors alone. Hence, the goal of thepresent study was to empirically test interac-tive effects of the physical and social environ-ment to gain a more comprehensive pictureof how the environment, broadly defined,contributes to childhood asthma.Within the literature on physical environ-ment, exposures such as traffic-related air pol-lution have been repeatedly linked to increasedrespiratory symptoms, increased asthma-related hospitalizations, and the diagnosis ofasthma (Brauer et al. 2007; Edwards et al.1994; Studnicka et al. 1997). In addition,experimental exposure to markers of traffic-related air pollution such as diesel exhaust par-ticles or nitrogen dioxide increases levels ofinflammatory markers relevant to asthma(Barck et al. 2002; DiazSanchez et al. 1997;Pourazar et al. 2004), suggesting that greaterinflammation may underlie the effects foundin clinical studies.Within the literature on social environ-ment, factors such as psychological stress havebeen linked to wheezing, asthma exacerba-tions, and the diagnosis of asthma in children(Klinnert et al. 2001; Sandberg et al. 2000;Wright et al. 2002). In addition, high levels ofstress have been associated with detrimentalbiologic profiles, such as greater inflammatoryresponses after antigen challenge or in vitrostimulation of immune cells, among childrenwith or at risk for asthma (Chen et al. 2006;Liu et al. 2002; Wright et al. 2004).Much of this research, although importantin its own right, has proceeded largely alongseparate lines, with little overlap. Few studieshave brought these two areas together toexamine how physical and social environ-ments might interact to affect asthma out-comes. Although it is possible that socialfactors alone could affect asthma outcomes, orthat physical factors would override any effectof social factors, a number of researchers haveargued that social and physical factors mightinstead combine to modify health outcomessuch as asthma (Weiss and Bellinger 2006).This interaction between the physical andsocial environments could mean that negativesocial factors intensify the effect of physicalenvironment exposures (Gee and Payne-Sturges 2004; Morello-Frosch and Shenassa2006). For example, the combination ofcertain physical and social environment riskfactors might create a “double exposure,”whereby the combination of factors such ashigh pollution exposure and high stress ismost detrimental to asthma. One recent study(Clougherty et al. 2007) empirically tested thenotion of physical by social environmentinteractions in asthma. This study involved alongitudinal investigation of a birth cohort ofchildren in which traffic-related air pollution(indicated by NO2) as well as chronic stress(indicated by exposure to violence) were meas-ured. The risk of asthma incidence was ele-vated as residential NO2 exposure increased,but only among children who had high levelsof exposure to violence (Clougherty et al.2007). In the present study, we sought todetermine whether a social by physical expo-sure interaction would predict outcomes inchildren with existing asthma. More specific-ally, we tested whether the physical and socialenvironment would interact to predict bio-logic markers implicated in asthma, and sub-sequently, whether their interaction wouldpredict changes in clinical outcomes overtime in a sample of children with asthma.Although previous research provides someevidence for asthma onset that suggests suchinteractions might fit a “double jeopardy”hypothesis, so few previous data address thisquestion in children with asthma that ourAddress correspondence to E. Chen, University ofBritish Columbia, Department of Psychology, 2136West Mall, Vancouver, BC V6T 1Z4 Canada.Telephone: (604) 822-2549. Fax: (604) 822-6923.E-mail: echen@psych.ubc.caThis study was supported by funding fromNational Institutes of Health grant HL073975 andthe Canadian Institutes of Health Research.The authors declare they have no competingfinancial interests.Received 15 November 2007; accepted 26 February2008.Chronic Traffic-Related Air Pollution and Stress Interact to Predict Biologicand Clinical Outcomes in AsthmaEdith Chen,1 Hannah M. C. Schreier,1 Robert C. Strunk,2 and Michael Brauer31Department of Psychology, University of British Columbia, Vancouver, British Columbia, Canada; 2Department of Pediatrics, Division ofAllergy and Pulmonary Medicine, St. Louis Children’s Hospital, Washington University in St. Louis School of Medicine, St. Louis,Missouri, USA; 3School of Environmental Health, University of British Columbia, Vancouver, British Columbia, CanadaBACKGROUND: Previous research has documented effects of both physical and social environmentalexposures on childhood asthma. However, few studies have considered how these two environ-ments might interact to affect asthma.OBJECTIVE: This study aimed to test interactions between chronic exposure to traffic-related air pollution and chronic family stress in predicting biologic and clinical outcomes in childrenwith asthma.METHOD: Children with asthma (n = 73, 9–18 years of age) were interviewed about life stress, andasthma-relevant inflammatory markers [cytokine production, immunoglobulin E (IgE), eosinophilcounts] were measured. Parents reported on children’s symptoms. Children completed daily diariesof symptoms and peak expiratory flow rate (PEFR) measures at baseline and 6 months later.Exposure to traffic-related air pollution was assessed using a land use regression model for nitrogendioxide concentrations.RESULTS: NO2 by stress interactions were found for interleukin-5 (β for interaction term = –0.31,p = 0.02), IgE (interaction β = –0.29, p = 0.02), and eosinophil counts (interaction β = –0.24,p = 0.04). These interactions showed that higher chronic stress was associated with heightenedinflammatory profiles as pollution levels decreased. Longitudinally, NO2 by stress interactionsemerged for daily diary symptoms (interaction β = –0.28, p = 0.02), parent-reported symptoms(interaction β = –0.25, p = 0.07), and PEFR (interaction β = 0.30, p = 0.03). These interactionsindicated that higher chronic stress was associated with increases over time in symptoms anddecreases over time in PEFR as pollution levels decreased.CONCLUSIONS: The physical and social environments interacted in predicting both biologic andclinical outcomes in children with asthma, suggesting that when pollution exposure is more mod-est, vulnerability to asthma exacerbations may be heightened in children with higher chronic stress.KEY WORDS: air pollution, asthma, immune, psychosocial, stress, traffic. Environ Health Perspect116:970–975 (2008). doi:10.1289/ehp.11076 available via http://dx.doi.org/ [Online 27 February2008]present study aimed to describe the nature ofany interaction patterns, rather than testingspecific a priori hypotheses.Materials and MethodsSubject. Seventy-three children were recruitedfrom Vancouver, British Columbia, Canada,through advertisements in physicians’ offices,local media, and community settings for anobservational study of childhood asthma.Advertisements were placed throughout theGreater Vancouver area, and interested familiescontacted the laboratory for a screening todetermine eligibility. Eligible families were thenscheduled for a laboratory visit. Visits occurredthroughout the year. Participants were requiredto be physician-diagnosed with asthma (82%with allergic asthma). Children were 9–18 yearsof age, fluent in English, free of acute respira-tory illness at the time of their visit (by parentand child report), and had no chronic illnessesother than asthma. Children gave writtenassent, and parents provided written consent.The protocol was approved by the University ofBritish Columbia Research Ethics Board.Traffic-related air pollution estimates. Aland use regression model developed for thestudy region (Henderson et al. 2007) providedhigh-resolution spatial estimates of NO2 con-centrations as an indicator of chronic exposureto traffic-related air pollution. Briefly, 116passive samplers to collect NO2 were deployedfor two 14-day periods at 116 sites in thestudy area. Mean concentrations during thesetwo periods were highly correlated with andclosely approximated annual averages fromregulatory monitoring network data.For each of the 116 measurement sites, 55variables were generated in a geographic infor-mation system, and a linear regression modelfor NO2 was built with the most predictivecovariates. For NO2, the model (R2 = 0.56)included the number of major roads within100- and 1,000-m radius circular buffers of themeasurement sites, the number of secondaryroads within a 100-m buffer, the populationdensity within a 2,500-m radius, the amountof commercial land use within 750 m, and ele-vation. Comparison of model predictions tomeasured annual average concentrations at 16government regulatory air-monitoring stationshad an R2 of 0.69. The predictive error for themodel was estimated with leave-one-out cross-validation, where each model is repeatedlyparametrized on n – 1 data points and thenused to predict the excluded measurement.The mean difference between predicted andmeasured values, an estimate of the modelerror, was 0.0 ppb with a standard deviation of2.75 ppb (~ 15% of the sample mean). The resulting surface was smoothed(ArcGis Spatial Analyst, Focal Statistics,Redlands, CA) to remove abrupt changes andedge effects to more accurately reflect themeasured effect of proximity to roadways(Gilbert et al. 2003). Using the model, we gen-erated a smooth spatial surface of predicted(annual average) concentrations for the entirestudy area at a resolution of 10 m. To adjustfor temporal variability, we fit the correspond-ing ambient monitoring network data withmonthly dummy variables and a covariate forlinear trend using the Times Series ForecastingSystem from SAS (version 9.1; SAS InstituteInc., Cary, NC, USA). We then applied thesemonth–year adjustment factors to the surfaceto estimate monthly average concentrations.Using these averages, we then computed indi-vidual subject average exposures for the fullperiod during which pollution estimates weretaken (1998–2003) as an indicator of chronictraffic-related air pollution. We included thefull period because the land use regressionmodel is best suited for long-term exposures(Marshall et al. 2008; Nethery et al. 2007), asit is based on spatial differences in land use thatdo not vary over time. Other approaches canbe used to assess short-term exposures but suf-fer from less spatial resolution. This specific airpollution exposure indicator has been associ-ated with bronchiolitis and asthma incidencein other studies in Vancouver (Clark et al.2007; Demers et al. 2006), and the same expo-sure assessment approach has been used inother cities to predict health outcomes such asasthma incidence, allergies, and wheezing(Brauer et al. 2002, 2007; Ryan et al. 2007).Psychosocial measure. We assessed chronicstress in children using the UCLA Life StressInterview (Hammen 1991). This interview wasconducted at baseline and assessed chronicstress over the preceding 6 months in domainssuch as family relationships, friendships, andschool. This is a semistructured interview inwhich a trained interviewer asks a series ofopen-ended questions in each life domain anduses the information gathered to rate the levelof chronic, ongoing stress. Interviewers rate theextent of chronic stress on a 1–5 scale, using0.5 increments, with higher numbers reflectingmore severe, persistent difficulties. This inter-view has been used successfully in children asyoung as 8 years of age and has demonstratedreliability and validity (Adrian and Hammen1993; Rudolph and Hammen 1999). In thepresent study we focused on chronic familystress because family stress, of all the life stressdomains, is the one most strongly related toasthma outcomes (Chen et al. 2006; Millerand Chen 2006).Immune measures. At the baseline visit,peripheral blood was drawn from childreninto BD vacutainer cell preparation tubes con-taining sodium heparin, and 3 × 106 periph-eral blood mononuclear cells (PBMCs) wereisolated through density-gradient centrifuga-tion. PBMCs were resuspended in culturemedium consisting of RPMI plus 10% fetalcalf serum and incubated with phorbol myris-tate acetate (25 ng/mL) and ionomycin(1 µg/mL) for 48 hr at 37°C in 5% carbondioxide . Supernatants were frozen until theend of the study and were then assayed todetermine levels of interleukin (IL)-4, IL-5,and IL-13 using enzyme-linked immunosor-bent assays (assays from R&D Systems,Minneapolis, MN, USA). Intraassay coeffi-cients of variation ranged from 3.68 to 4.76%.We performed a complete blood countwith five-part differential (Bayer ADVIA 70hematology system; Holiston, MA, USA) toobtain eosinophil counts. Total serumimmunoglobulin E (IgE) was measured usingan automated fluorescence immunoassay(Pharmacia CAP system, Portage, MI, USA),and was log transformed because of its non-normal distribution.Clinical measures. To obtain multiple per-spectives on asthma symptoms, both parentsand children were probed about symptoms.Parents were interviewed during the laboratoryvisit about their child’s symptoms over the pre-vious 2 weeks as part of a health interview thatgathered information about the child’s asthma(e.g., prescribed medications). The frequency ofdaytime symptoms, nighttime symptoms, andexertional symptoms were probed according tothe National Asthma Education and PreventionProgram Expert Panel Report 2 (NAEPP/EPR2) guidelines (NAEPP 1997). We assesseddaytime symptoms as the number of days overthe previous 2 weeks children had cough,wheeze, shortness of breath, or chest tightness.We assessed nighttime symptoms as the num-ber of days children awakened from sleepbecause of coughing, wheezing, shortness ofbreath, or chest tightness. And we assessed exer-tional symptoms as the number of days chil-dren had cough, wheeze, shortness of breath, orchest tightness while exercising or playing.Children were asked to rate their symp-toms on a daily basis for 2 weeks after theirlaboratory visit. Children were asked to keepa diary of their symptoms, and every morningand evening they rated the extent of cough-ing, wheezing, chest tightness, and shortnessof breath, each on a 0 (none) to 4 (really bad)scale. Scores for each question were summedfor each day, then averaged across the 14-daydiary period.Over the same period, children also moni-tored peak expiratory flow rate (PEFR) at homeusing an electronic monitor (Quadromed,Hoechberg, Germany). Three peak flowreadings were taken on awakening and beforebedtime each day for 2 weeks, and the highestvalue at each time point was retained. DailyPEFR% was calculated as a percent of eachyouth’s laboratory best, and readings across the2 weeks were averaged.All clinical measures were repeated6 months after the baseline visit.Stress and air pollution in asthmaEnvironmental Health Perspectives • VOLUME 116 | NUMBER 7 | July 2008 971Potential confounders. Variables thatcould provide alternative explanations for theabove relationships were included as covariatesin statistical analyses. This included asthmaseverity, determined from the NAEPP/EPR2guidelines based on the higher of symptomfrequency and medication use, paralleling theapproach of previous researchers (Bacharieret al. 2004). Families also brought children’sasthma medications to the research center, andinhaled corticosteroid use was coded (numberof days taken during the preceding 2 weeks),as was beta agonist use (number of days takenduring the preceding 2 weeks).In addition, we assessed whether demo-graphic characteristics or study visit variables,such as child age, sex, ethnicity, or time ofyear of study visit were associated with studyvariables. Demographic or study variables thatwere significantly associated with anyimmune or clinical measures were included ascovariates in analyses with that outcome.Statistical analyses. We conducted statisti-cal analyses to test the hypothesis that psycho-logic stress would interact with air pollutionexposure to predict both biologic and clinicalasthma outcomes. Biologic variables wereexamined cross-sectionally and clinical out-comes longitudinally, using a series of hierar-chical multiple regression analyses. Biologicvariables were predicted from variables enteredin three steps: a) potential confounders includ-ing medical variables (asthma severity classifi-cation, use of inhaled corticosteroids, use ofbeta agonists) and any demographic variableassociated with study outcomes; b) maineffects of chronic stress and air pollution; andc) the interaction term for chronic stress by airpollution. These analyses were conductedaccording to the recommendations by Aikenand West (1991). Because both chronic stressand pollution exposure are continuous vari-ables, these involve statistical procedures toexamine the significance of an interactionbetween two continuous variables. Theseanalyses comprised the primary test of studyhypotheses. The nature of an interactionbetween two continuous variables, however,can be difficult to visualize. To aid in interpre-tation, we provided two additional pieces ofinformation. Significant interactions wereplotted by graphing the relationship betweenstress and asthma outcomes at low (–1 SD)and high (+1 SD) levels of air pollution. Thiscreates an artificial distinction within one ofthe continuous variables, but allows one tomore easily see how the relationship betweenstress and asthma varies at different levels ofpollution exposure. We also conducted sec-ondary analyses to test whether the regressioncoefficients at these specific values of air pollu-tion were significant. Again, this creates anartificial distinction, but allows the reader togain some sense of where the differences aremost pronounced. These procedures followthe statistical recommendations of Aiken andWest (1991). However, because these analysesdo not fully represent the nature of the studyvariables (as continuous), they are consideredsecondary, with the test of the interactionbetween the two continuous variables as theprimary analyses.Clinical variables included two timepoints spaced 6 months apart. In these analy-ses, the difference score (time 2 – time 1) waspredicted from the same set of variablesdescribed above, except that time 1 valueswere included as a control variable in step 1.ResultsTable 1 presents descriptive information aboutthe sample. There were a few associations ofdemographic and study variables with biologicor clinical outcomes. Child age was inverselycorrelated with eosinophil count (r = –0.23,p = 0.04), positively correlated with daily diarysymptoms (r = 0.25, p = 0.04), and positivelycorrelated with PEFR% (r = 0.24, p = 0.04).Parents reported girls to have more symptomsthan boys (t = 3.36, p = 0.001). Childrenbelonging to minority groups had higher pro-duction of IL-4 compared with white children(t = 3.40, p = 0.001). Time of year was cor-related with IL-13 production (r = –0.22,p = 0.05), such that higher levels were foundfor those children seen during the wintermonths. Variables associated with outcomeswere included as covariates in the relevantequations below.Cross-sectional associations with biologicmarkers. Our first set of analyses testedwhether stress and air pollution were associ-ated with biologic markers implicated inasthma. Regression coefficients are reportedbelow, with interactions graphed to illustratethe direction of effects.With respect to IL-5 production, there wasno main effect of stress (β = 0.16, p = 0.22) orair pollution (β = 0.15, p = 0.30), but there wasa significant stress by air pollution interaction(interaction term β = –0.31, p = 0.02). Thenegative coefficient indicates that as pollutiondecreases, higher levels of chronic family stressbecome associated with greater production ofIL-5. This is illustrated in Figure 1A by plottingthe relationship between chronic stress and IL-5production at low and high levels of air pollu-tion. The relationship between stress and IL-5production was positive and statistically signifi-cant at 1 SD below the mean of air pollution(β = 0.39, p = 0.03), whereas the relationshipbetween stress and IL-5 production at 1 SDabove the mean of air pollution was not signifi-cant (β = –0.06, p = 0.71), suggesting that therelationship between stress and IL-5 productionis stronger in lower-pollution areas.No significant associations emerged forIL-4 or IL-13 (p > 0.2).With respect to total IgE, there was asignificant main effect of stress such thatChen et al.972 VOLUME 116 | NUMBER 7 | July 2008 • Environmental Health PerspectivesTable 1. Descriptive information on study participants.Characteristic ValueAge 12.82 ± 2.75Sex (%)Male 68Female 32Ethnicity (%)White 63Asian 26Other 11Severity (%)Mild intermittent 16Mild persistent 38Moderate persistent 32Severe persistent 14Inhaled corticosteroidsa 4.35 ± 5.91Beta agonistsa 3.93 ± 5.55Chronic stressb 2.14 ± 0.74NO2 (ppb) 16.5 ± 3.7IL-4 (pg/mL) 12.47 ± 10.85IL-5 (pg/mL) 110.46 ± 91.00IL-13 (pg/mL) 323.42 ± 237.22IgE (log-transformed kU/L) 2.18 ± 0.80Eosinophil count (× 109cells/L) 0.36 ± 0.28Parent-reported symptomscBaseline 1.96 ± 2.796-month follow-up 1.23 ± 2.08Child-reported daily diary symptomsdBaseline 3.42 ± 3.856-month follow-up 2.61 ± 3.23Daily peak expiratory flow rate (%)Baseline 99.83 ± 15.626-month follow-up 95.08 ± 16.64Values are mean ± SD except where indicated.aMedication values are for number of days taken in thepreceding 2 weeks. bChronic stress is on a 1–5 scale.cParent-reported symptoms = average number of days ofsymptoms in preceding 2 weeks reported during the labo-ratory visit. dChild-reported daily diary symptoms = averagedaily symptom score from the 2-week home monitoringafter the laboratory visit. Figure 1. (A) Interaction between chronic stress and air pollution predicting IL-5 production. The graphdisplays the estimated regression line for the relationship between chronic stress and IL-5 production atlow (–1 SD) and high (+1 SD) levels of air pollution. (B) Interaction between chronic stress and air pollutionfor total IgE. (C) Interaction between chronic stress and air pollution for eosinophil counts.1501209060300Low HighChronic family stressIL-5 (pg/mL)Low HighChronic family stressTotal IgE (Iog10)2.52.21.91.61.31.0Low HighChronic family stress0.70.60.50.40.30.2Eosinophil countA B CLow pollutionHigh pollutionchildren with higher chronic stress had higherIgE levels (β = 0.32, p = 0.01), but no maineffect of air pollution (β = 0.08, p = 0.52). Inaddition, there was a significant stress by airpollution interaction (interaction term β =–0.29, p = 0.02). The negative coefficient indi-cates that as pollution decreases, higher levels ofchronic family stress become associated withgreater IgE levels. This is illustrated in Figure1B by plotting the relationship between chronicstress and IgE at +1 SD and –1 SD of air pollu-tion. The relationship between stress and IgEwas positive and statistically significant at 1 SDbelow the mean of air pollution (β = 0.54, p =0.002), whereas the relationship between stressand IgE at 1 SD above the mean of air pollu-tion was not significant (β = 0.11, p = 0.46),suggesting that the relationship between stressand IgE is stronger in lower pollution areas.With respect to eosinophil counts, therewas no main effect of stress (β = 0.05, p =0.67), and a weak effect of air pollution (β =0.21, p = 0.10). There was a significant stressby air pollution interaction (interaction term β= –0.24, p = 0.04). The negative coefficientindicates that as pollution decreases, higher lev-els of chronic family stress become associatedwith greater eosinophil counts. This is illus-trated in Figure 1C by plotting the relationshipbetween chronic stress and eosinophil counts at+1 SD and –1 SD of air pollution. The rela-tionship between stress and eosinophil countswas positive at 1 SD below the mean of air pol-lution (β = 0.23, p = 0.15), whereas the rela-tionship between stress and eosinophil countsat 1 SD above the mean of air pollution wasnegative (β = –0.13, p = 0.35).Longitudinal associations with clinical out-comes. Table 2 presents descriptive informationon clinical variables at baseline and follow-upon the sample. This information is separatedinto a low- and a high-pollution group bymedian split so readers can see how clinicalvariables vary by pollution. Average (± SD)pollution exposure for those below the medianwas 14.1 ± 1.6 ppb. Average exposure for thoseabove the median was 18.9 ± 3.5 ppb. Low-and high-pollution groups were similar ondemographic variables such as age (average agefor those below the median on pollution expo-sure = 12.2 years; average age for those abovethe median = 13.5 years), ethnicity (66% whiteamong those below the median on pollutionexposure; 61% white among those above themedian), and parent education (average yearsof education for those below the median onpollution exposure = 15.4; average years of edu-cation for those above the median = 15.6). Thetable reveals that children in high pollutionareas had higher child-reported daily symptomsand parent-reported symptoms at both timepoints than those in low-pollution areas. In theanalyses below, we focused on change overtime in clinical variables to assess whether thecross-sectional associations with biologic mark-ers have implications clinically for asthma overtime. To do this, we tested whether the interac-tion between stress and air pollution predictedchanges in clinical variables over a 6-monthperiod, controlling for baseline levels.With respect to the daily diaries that chil-dren kept of symptoms, there was no maineffect of stress (β = 0.06, p = 0.63) or air pol-lution (β = –0.12, p = 0.38), but there was asignificant stress by air pollution interaction(interaction term β = –0.28, p = 0.02). Thenegative coefficient indicates that as pollutiondecreases, higher levels of chronic family stressbecome associated with increasing symptomsover time. This is illustrated in Figure 2A byplotting the relationship between chronicstress and change in symptoms at +1 SD and–1 SD of air pollution. The relationshipbetween stress and symptom change was posi-tive at 1 SD below the mean of air pollution(β = 0.25, p = 0.098), whereas the relation-ship between stress and symptom change at1 SD above the mean of air pollution wasnegative (β = –0.13, p = 0.35).With respect to parent report of childsymptoms, there was no main effect of stress(β = 0.08, p = 0.60) or air pollution (β = –0.03,p = 0.85), but there was a marginal stress by airpollution interaction (β = –0.25, p = 0.07). Thenegative coefficient indicates that as pollutiondecreases, higher levels of chronic family stressbecome associated with increasing parent-reported symptoms over time. This is illustratedin Figure 2B by plotting the relationshipbetween chronic stress and change in symptomsat +1 SD and –1 SD of air pollution. The rela-tionship between stress and symptom changewas positive at 1 SD below the mean of airpollution (β = 0.36, p = 0.08), whereas the rela-tionship between stress and symptom change at1 SD above the mean of air pollution was nega-tive (β = –0.21, p = 0.36).With respect to daily PEFR measures,there was no main effect of stress (β = 0.05, p =0.68) or air pollution (β = 0.06, p = 0.70), butthere was a significant stress by air pollutioninteraction (β = 0.30, p = 0.03). The positivecoefficient indicates that as pollution decreases,higher levels of chronic family stress alsobecome associated with decreasing PEFR overtime. This is illustrated in Figure 2C by plot-ting the relationship between chronic stressand change in PEFR at +1 SD and –1 SD ofair pollution. The relationship between stressand PEFR change was negative at 1 SD belowthe mean of air pollution (β = –0.14, p = 0.40),whereas the relationship between stress andPEFR change at 1 SD above the mean of airpollution was positive (β = 0.24, p = 0.11).Although the direction of change differedby pollution levels, this does not meanthat symptoms are actually higher in lower-pollution areas. As shown in Table 2, childrenabove the median in pollution exposure hadgreater symptoms by daily diary report andparent report than did children below themedian in pollution exposure, and had com-parable PEFRs. Hence, children in higher-pollution areas have greater symptoms, butthese symptoms do not appear to worsen overtime the way they do for children in lower-pollution areas with chronic stress.DiscussionThis article is the first that we are aware of todocument that physical environment (chronictraffic-related air pollution) and socialStress and air pollution in asthmaEnvironmental Health Perspectives • VOLUME 116 | NUMBER 7 | July 2008 973Table 2. Descriptive information on clinical measures by pollution group (mean ± SD).Low pollutiona High pollutionbChild daily diary symptoms, baseline 2.79 ± 3.42 4.08 ± 4.21Child daily diary symptoms, follow-up 2.09 ± 2.35 3.16 ± 3.90Parent-reported symptoms, baseline 3.47 ± 4.44 8.47 ± 10.67Parent-reported symptoms, follow-up 2.91 ± 6.22 4.50 ± 6.26PEFR%, baseline 100.02 ± 16.07 99.62 ± 15.35PEFR%, follow-up 94.15 ± 12.93 96.07 ± 20.04aThose below the median on NO2 scores. bThose above the median on NO2 scores.Figure 2. (A) Interaction between chronic stress and air pollution predicting change in child-reported dailydiaries of asthma symptoms over a 6-month period. The graph displays the estimated regression line forthe relationship between chronic stress and changes in asthma symptoms over time at low (–1 SD) andhigh (+1 SD) levels of air pollution. (B) Interaction between chronic stress and air pollution for changes inparent-reported symptoms over a 6-month period. (C) Interaction between chronic stress and air pollutionfor changes in daily PEFR percent over a 6-month period.6420–2Low HighChronic family stressChange in child-reported daily diarysymptoms over 6 monthsLow HighChronic family stressLow HighChronic family stress181260–6–12Change in daily PEFR %over 6 monthsA B CLow pollutionHigh pollution6420–2Change in parent-reportedsymptoms over 6 monthsenvironment (chronic stress) interact to predictboth biologic and longitudinal clinical out-comes in children with asthma. The findingsfrom this study demonstrated that the inter-active effects between air pollution and stressare stronger than either factor alone, suggestingthat the physical and social environments arein fact intertwined and critical to understandin concert, rather than independently.The nature of this interaction was suchthat the detrimental effects of chronic psy-chosocial stress were more evident amongchildren living in lower-pollution areas. Thatis, as pollution levels declined, higher levels ofstress were associated with heightened inflam-matory profiles cross-sectionally and worsen-ing clinical profiles over a 6-month period. Incontrast, chronic stress had modest effects onbiologic and clinical measures as pollutionexposure increased, and any suggestions ofeffects were in an opposite direction.The direction of the interaction effects inthis study was different from that found in thesmall number of previous studies on this topic.For example, one recent study found thattraffic-related air pollution (NO2) interactedwith exposure to violence to predict the diagno-sis of asthma in a birth cohort of children, suchthat children with both high pollution and vio-lence exposures were at greatest risk of havingasthma (Clougherty et al. 2007). Another studyfound that high levels of traffic-related air pol-lution (e.g., NO2) combined with low socioe-conomic status predicted the greatest risk ofasthma hospitalizations in children (Lin et al.2004). These previous studies fit a “doublejeopardy” hypothesis, suggesting that the com-bination of physical and social exposures syner-gistically affect asthma outcomes.In contrast, our data suggest that chronicstress may have the ability to accentuate theeffects of environmental pollutants whenchronic exposure levels are more modest.Previous research has suggested that social fac-tors do require the presence of some dose ofphysical exposure to have effects on biologicprocesses (Chen and Miller 2007). The find-ings from the present study fit the notion of athreshold model—that is, that there is athreshold at which chronic physical exposuresbegin to have effects on health outcomes, andthat one role of chronic stress may be to lowerthe threshold at which physical exposures affectbiologic and clinical outcomes. One reasonwhy this may occur is that when chronic expo-sure to traffic-related air pollutants is moremodest, there may be greater room for socialfactors to increase or decrease vulnerability bio-logically. This is consistent with the notionthat stress may be able to shift physiologicresponse systems in a direction such thatadverse outcomes occur in response to lowerdoses of physical exposures (Morello-Froschet al. 2006; Paarlberg et al. 1995). Consistentwith this notion, chronic stress under certainconditions has been found to heighten biologicresponses to negative social exposures (Gumpand Matthews 1999); we speculate that similarprocesses may occur with responses to physicalexposures. This type of response pattern isthought to occur because prolonged stress cansensitize and prevent adaptation of biologicsystems (McEwen 1998), potentially leading tolower doses of physical environment pollutantshaving detrimental effects on biologic andclinical asthma measures.Reasons for the differences between ourstudy and the two studies cited above areunclear, but we speculate that they may beattributable to distinctions between the diag-nosis versus progression of a disease, or to dif-ferent conceptualizations of social exposures.Our study focused on children with preexistingasthma and predicted biologic outcomes aswell as changes in clinical outcomes over time;in contrast, the study by Clougherty et al.(2007) predicted the risk of being diagnosedwith asthma. The way in which physical andsocial exposures affect the onset versus progres-sion of disease could be different, resulting inthe distinct patterns found in the two studies.In addition, different types of social exposuresmay have different effects on asthma. Althoughthey all are forms of stress, exposure to vio-lence, chronic family stress, and low socioeco-nomic status each represents different types oflife stressors, and it is possible that air pollutioninteracts differently with different types ofstressors. In the present study, we focused onfamily stress because this type of stress has themost robust associations with asthma out-comes (Chen et al. 2006; Miller and Chen2006); nonetheless, it is possible that otherunmeasured stressors, such as living in impov-erished neighborhoods, also contribute toasthma biologic and clinical outcomes andoverlap with air pollution indicators. Futureresearch should test these possibilities further.The effects that we found cross-sectionallyof interactions between chronic traffic-relatedair pollution and chronic stress on IL-5 pro-duction, total IgE levels, and eosinophilcounts represent biologic pathways that haveimplications for clinical asthma outcomes.Immune pathways in asthma include thesecretion of cytokines that activate B cells toproduce IgE. IgE in turn initiates an inflam-matory cascade leading to airway constrictionand mucus production. A second pathwayinvolves the recruitment of eosinophils to theairways, which also promotes airway inflam-mation and obstruction. Secretion of thecytokine IL-5 is known to increase eosinophilproduction. Thus heightened production ofIL-5 along with elevated IgE and eosinophilcounts suggests a biologic profile that is poten-tially detrimental for children with asthma interms of vulnerability to symptoms.We considered this possibility by testingwhether chronic traffic-related air pollutionby chronic stress interactions could also pre-dict changes in clinical outcomes over a6-month period. Consistent with the implica-tions of the cross-sectional biologic data, wefound that in lower-pollution areas, higherlevels of chronic stress at baseline predictedincreases in asthma symptoms and decreasesin daily PEFR over time. Hence chronic stressappears to exacerbate the effects of moremodest exposures to chronic air pollutants onlonger-term clinical asthma outcomes, inaddition to biologic markers.Interestingly, as pollution levels increased,an opposite pattern emerged whereby higherlevels of chronic stress were associated withdeclines in asthma symptoms and increases inPEFR over time. This was an unexpected trend,and it is unclear what the implications are.However, because longitudinal analyses focuson change over time, this does not mean thatchildren in higher-pollution areas have absolutelevels of asthma morbidity that are low. Rather,children in higher-pollution areas have moresymptoms than children in lower-pollutionareas at both time points, but their clinical pro-files (as assessed by PEFR and symptom report-ing) do not appear to worsen over time. Incontrast, children in lower-pollution areas showstronger relationships of chronic stress withworsening clinical profiles over time.Strengths of the present study include thecollection of asthma-relevant biomarkers; thetracking of longitudinal clinical outcomes; theuse of a land-use regression model to assessindividual exposures to air pollution; and theuse of an in-depth interview for measuringchronic stress. In addition, the design of thestudy meant that directionality could be moreclearly inferred. For example, although it wouldbe reasonable to hypothesize that more severeasthma increases family stress levels, the factthat stress was assessed before clinical measuresfor longitudinal analyses meant that worseningasthma was not driving stress experiences.Limitations to the present study includethe small sample size. Both the comprehensivestress interview and the collection and process-ing of biologic samples limited the size of thepresent sample, and this raises the possibilitythat findings may have been attributed tochance and hence need to be replicated.However, our sample size is comparable tothose of numerous other studies of acute stressand asthma cytokine production (Kang andFox 2001; Kang et al. 1997; Marshall et al.1998). A second limitation is the varying timeframe for measures in this study. Time frameswere set based on optimal periods for gatheringinformation for different study constructs. Forexample, symptom reports are best assessedover shorter intervals (weeks), given the diffi-culties in accurately recalling symptoms overChen et al.974 VOLUME 116 | NUMBER 7 | July 2008 • Environmental Health PerspectivesStress and air pollution in asthmaEnvironmental Health Perspectives • VOLUME 116 | NUMBER 7 | July 2008 975longer time periods (NAEPP 1997). Chronicstress is best assessed over a period of monthsto accurately capture persistent stressful influ-ences in different life domains (Hammen1991). Finally, pollution estimates using landuse regression models are best suited for long-term exposures, given that the model is basedon spatial differences in land use that do notvary over time. Because pollution data wereavailable for a 6-year period (1998–2003), butthis period did not overlap with the timeframes of the other study constructs, we usedthe entire period as a more reliable indicator oflong-term exposures. One limitation of thisapproach is that if families moved, the esti-mated exposures would be misclassified; how-ever, this increase in measurement error wouldbe expected to decrease the likelihood ofobserving associations. Nonetheless, futurestudies that a) are able to more precisely coor-dinate the periods of assessment for air pollu-tion, stress, and clinical outcomes, and b)could repeatedly assess families to trackchanges in pollution exposures based onmoves, as well as changes in chronic stressexperiences over time, would be useful formore clearly delineating the time frame ofeffects of physical and social exposures. A thirdlimitation is the lack of health records to ascer-tain objective asthma-relevant outcomes suchas hospitalizations and physician visits. Futurestudies that have access to such databaseswould allow researchers to explore additionalclinical indicators that may be influenced byboth the physical and social environments.Finally, we used NO2 as an indicator of traffic-related air pollution, and effects may be attrib-utable specifically to NO2. Further, althoughwe assessed pollution exposure with a high-res-olution spatial model estimating air pollutionconcentrations at the individual subjects’ homeaddress, we did not consider short-term tem-poral variability in exposures during the studyperiod, nor did we consider other environ-ments (e.g., schools) where participants spenda good deal of time. Because it was not feasibleto measure the actual level of pollution expo-sure that each child experienced throughoutthe day, this raises the possibility that some ofthe patterns could have been affected byunmeasured exposures, particularly in a smallsample such as this one.In summary, in this study we found thatthe physical and social environment interact toaffect asthma outcomes in children. As pollu-tion levels declined, higher levels of chronicfamily stress were associated with heightenedinflammatory profiles cross-sectionally and withincreases in asthma symptoms and decreases inpeak expiratory flow over a 6-month period.Conversely, as pollution levels increased,chronic stress either had no effect on outcomes(inflammatory measures) or in some casesshowed an opposite effect (clinical measures).These findings suggest that vulnerability factorssuch as psychosocial stress most clearly modifythe effects of traffic-related air pollution whenexposure is present but not high. This high-lights the need for families who have a childwith asthma to be increasingly aware of howthe physical and social environments are inter-twined, and how vulnerability to asthma exac-erbations may persist in children who areexperiencing higher levels of chronic stress, evenin areas with lower levels of pollution.REFERENCESAdrian C , Hammen C. 1993. Stress exposure and stress gener-ation in children of depressed mothers. J Consult ClinPsychol 61:354–359.Aiken LS , West SG. 1991. Multiple Regression: Testing andInterpreting Interactions. London:Sage Publications.Bacharier LB, Strunk RC, Mauger D, White D, Lemanske RF,Sorkness CA. 2004. Classifying asthma severity in children:mismatch between symptoms, medication use, and lungfunction. Am J Respir Crit Care Med 170:426–432.Barck C, Sandstrom T, Lundahl J, Hallden G, Svartengren M,Strand V, et al. 2002. Ambient level of NO2 augments theinflammatory response to inhaled allergen in asthmatics.Respir Med 96:907–917.Brauer M, Hoek G, Smit HA, de Jongste JC, Gerritsen J,Postma DS, et al. 2007. Air pollution and development ofasthma, allergy and infections in a birth cohort. Eur RespirJ 29:879–888.Brauer M, Hoek G, Van Vliet P, Meliefste K, Fischer PH, Wijga A,et al. 2002. Air pollution from traffic and the development ofrespiratory infections and asthmatic and allergic symptomsin children. Am J Respir Crit Care Med 166:1092–1098.Chen E, Hanson MD, Paterson LQ, Griffin MJ, Walker HA,Miller GE. 2006. Socioeconomic status and inflammatoryprocesses in childhood asthma: the role of psychologicalstress. J Allergy Clin Immunol 117:1014–1020.Chen E, Miller GE. 2007. Stress and inflammation in exacerba-tions of asthma. Brain Behav Immun 21:993–999.Clark N, Demers P, Karr C, Koehoom M, Tamburic L, Brauer M.2007. Effect of ambient air pollution exposure on develop-ment of childhood asthma. Epidemiology 18:S180–S181.Clougherty JE, Levy JI, Kubzansky LD, Ryan PB, Suglia SF,Jacobson Canner M, et al. 2007. Synergistic effects of traf-fic-related air pollution and exposure to violence on urbanasthma etiology. Environ Health Perspect 115:1140–1146.Crain EF, Walter M, O’Connor GT, Mitchell H, Gruchalla RS,Kattan M, et al. 2002. Home and allergic characteristics ofchildren with asthma in seven U.S. urban communities anddesign of an environmental intervention: The Inner-CityAsthma Study. Environ Health Perspect 110:939–945.Demers P, Karr C, Koehoom M, Lencar C, Tamburic L, BrauerM. 2006. The impact of air pollution on bronchiolitis.Epidemiology 17:S147.DiazSanchez D, Tsien A, Fleming J, Saxon A. 1997. Combineddiesel exhaust particulate and ragweed allergen chal-lenge markedly enhances human in vivo nasal ragweed-specific IgE and skews cytokine production to a T helpercell 2-type pattern. J Immunol 158:2406–2413.Edwards J, Walters S, Griffiths RK. 1994. Hospital admissionsfor asthma in preschool children: relationship to majorroads in Birmingham, United Kingdom. Arch EnvironHealth 49:223–227.Gee GC, Payne-Sturges DC. 2004. Environmental health dispari-ties: a framework integrating psychosocial and environ-mental concepts. Environ Health Perspect 112:1645–1653.Gilbert NL, Woodhouse S, Stieb DM, Brook JR. 2003. Ambientnitrogen dioxide and distance from a major highway. SciTotal Environ 312:43–46.Gump BB, Matthews KA. 1999. Do background stressors influ-ence reactivity to and recovery from acute stressors?J Appl Soc Psychol 29:469–494.Hammen C. 1991. Generation of stress in the course of unipolardepression. J Abnorm Psychol 100:555–561.Henderson SB, Beckerman B, Jerrett M, Brauer M. 2007.Application of land use regression to estimate long-termconcentrations of traffic-related nitrogen oxides and fineparticulate matter. Environ Sci Technol 41:2422–2428.Kang D, Coe C, McCarthy DO, Jarjour NN, Kelly EA, RodriguezRR, et al. 1997. Cytokine profiles of stimulated blood lym-phocytes in asthmatic and healthy adolescents across theschool year. J Interferon Cytokine Res 17:481–487.Kang DH, Fox C. 2001. Th1 and Th2 cytokine responses to acad-emic stress. Res Nurs Health 24:245–257.Klinnert MD, Nelson HS, Price MR, Adinoff AD, Leung DY,Mrazek DA. 2001. Onset and persistence of childhoodasthma: predictors from infancy. Pediatrics 108:e69.Lin M, Chen Y, Villeneuve PJ, Burnett RT, Lemyre L, HertzmanC, et al. 2004. Gaseous air pollutants and asthma hospital-ization of children with low household income inVancouver, British Columbia, Canada. Am J Epidemiol159:294–303.Liu LY, Coe CL, Swenson CA, Kelly EA, Kita H, Busse WW. 2002.School examinations enhance airway inflammation to anti-gen challenge. Am J Respir Crit Care Med 165:1062–1067.Marshall GD, Agarwal SK, Lloyd C, Cohen L, Henninger EM,Morris GJ. 1998. Cytokine dysregulation associated withexam stress in healthy medical students. Brain BehavImmun 12:297–307.Marshall JD, Nethery E, Brauer M. 2008. Within-urban variabil-ity in ambient air pollution: comparison of estimation meth-ods. Atmos Environ 42(6):1359–1369.McEwen BS. 1998. Protective and damaging effects of stressmediators. N Engl J Med 338:171–179.Miller GE, Chen E. 2006. Life stress and diminished expressionof genes encoding glucocorticoid receptor and beta(2)-adrenergic receptor in children with asthma. Proc NatlAcad Sci USA 103:5496–5501.Morello-Frosch R, Shenassa ED. 2006. The environmental“riskscape” and social inequality: implications for explain-ing maternal and child health disparities. Environ HealthPerspect 114:1150–1153.NAEPP (National Asthma Education and Prevention Program).1997. Expert Panel Report 2: Guidelines for the Diagnosisand Management of Asthma. NIH publication no. 97–4051.Bethesda, MD:National Institutes of Health.Nethery E, Leckie SE, Teschke K, Brauer M. 2007. From meas-ures to models: an evaluation of air pollution exposureassessment for epidemiologic studies of pregnant women.Occup Environ Med; doi:10.1136/oem.2007.035337 [Online10 December 2007]. Paarlberg KM, Vingerhoets AJJM, Passchier J, Dekker GA,Vangeijn HP. 1995. Psychosocial factors and pregnancyoutcome: a review with emphasis on methodologicalissues. J Psychosom Res 39:563–595.Pourazar J, Frew AJ, Blomberg A, Helleday R, Kelly FJ, WilsonS, et al. 2004. Diesel exhaust exposure enhances theexpression of IL-13 in the bronchial epithelium of healthysubjects. Respir Med 98:821–825.Rudolph KD, Hammen C. 1999. Age and gender as determinantsof stress exposure, generation, and reactions in young-sters: a transactional perspective. Child Dev 70:660–677.Ryan PH, LeMasters GK, Biswas P, Levin L, Hu SH, Lindsey M,et al. 2007. A comparison of proximity and land use regres-sion traffic exposure models and wheezing in infants.Environ Health Perspect 115:278–284.Sandberg S, Paton JY, Ahola S, McCann DC, McGuinness D,Hillary CR. 2000. The role of acute and chronic stress inasthma attacks in children. Lancet 356:982–987.Studnicka M, Hackl E, Pischinger J, Fangmeyer C, Haschke N,Kuhr J, et al. 1997. Traffic-related NO2 and the prevalenceof asthma and respiratory symptoms in seven year olds.Eur Respir J 10:2275–2278.Weiss B, Bellinger DC. 2006. Social ecology of children’s vul-nerability to environmental pollutants. Environ HealthPerspect 114:1479–1485.Wright RJ. 2005. Stress and atopic disorders. J Allergy ClinImmunol 116:1301–1306.Wright RJ, Cohen S, Carey V, Weiss ST, Gold DR. 2002.Parental stress as a predictor of wheezing in infancy: aprospective birth-cohort study. Am J Respir Crit Care Med165:358–365.Wright RJ, Finn P, Contreras JP, Cohen S, Wright RO,Staudenmayer J, et al. 2004. Chronic caregiver stress andIgE expression, allergen-induced proliferation, andcytokine profiles in a birth cohort predisposed to atopy.J Allergy Clin Immunol 113:1051–1057.

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.52383.1-0220725/manifest

Comment

Related Items