UBC Faculty Research and Publications

A new exposure metric for traffic-related air pollution? An analysis of determinants of hopanes in settled… Sbihi, Hind; Brook, Jeffrey R; Allen, Ryan W; Curran, Jason H; Dell, Sharon; Mandhane, Piush; Scott, James A; Sears, Malcolm R; Subbarao, Padmaja; Takaro, Timothy K; Turvey, Stuart E; Wheeler, Amanda J; Brauer, Michael Jun 19, 2013

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

Item Metadata

Download

Media
52383-12940_2013_Article_665.pdf [ 280.65kB ]
Metadata
JSON: 52383-1.0223294.json
JSON-LD: 52383-1.0223294-ld.json
RDF/XML (Pretty): 52383-1.0223294-rdf.xml
RDF/JSON: 52383-1.0223294-rdf.json
Turtle: 52383-1.0223294-turtle.txt
N-Triples: 52383-1.0223294-rdf-ntriples.txt
Original Record: 52383-1.0223294-source.json
Full Text
52383-1.0223294-fulltext.txt
Citation
52383-1.0223294.ris

Full Text

RESEARCH Open AccessA new exposure metric for traffic-related airin settled indoor house dustHind Sbihi1*, Jeffrey R Brook2, Ryan W Allen3, Jason H Curran1, Sharon Dell4, Piush Mandhane5, James A Scott6,Malcolm R Sears7, Padmaja Subbarao4, Timothy K Takaro3, Stuart E Turvey8, Amanda J Wheeler9and Michael Brauer1AbstractBackground: Exposure to traffic-related air pollution (TRAP) can adversely impact health but epidemiologic studiesare limited in their abilities to assess long-term exposures and incorporate variability in indoor pollutant infiltration.Methods: In order to examine settled house dust levels of hopanes, engine lubricating oil byproducts found invehicle exhaust, as a novel TRAP exposure measure, dust samples were collected from 171 homes in five Canadiancities and analyzed by gas chromatography–mass spectrometry. To evaluate source contributions, the relativeabundance of the highest concentration hopane monomer in house dust was compared to that in outdoor air.Geographic variables related to TRAP emissions and outdoor NO2 concentrations from city-specific TRAP land useregression (LUR) models were calculated at each georeferenced residence location and assessed as predictors ofvariability in dust hopanes.Results: Hopanes relative abundance in house dust and ambient air were significantly correlated (Pearson’s r=0.48,p<0.05), suggesting that dust hopanes likely result from traffic emissions. The proportion of variance in dusthopanes concentrations explained by LUR NO2 was less than 10% in Vancouver, Winnipeg and Toronto while thecorrelations in Edmonton and Windsor explained 20 to 40% of the variance. Modeling with household factors suchas air conditioning and shoe removal along with geographic predictors related to TRAP generally increased theproportion of explained variability (10-80%) in measured indoor hopanes dust levels.Conclusions: Hopanes can consistently be detected in house dust and may be a useful tracer of TRAP exposure ifdeterminants of their spatiotemporal variability are well-characterized, and when home-specific factors are considered.Keywords: Air pollution, Dust, Exposure assessment, Hopanes, Land use regression, TrafficBackgroundExposure to traffic-related air pollutants (TRAP) is associ-ated with excess mortality [1,2]. The burden of air pollu-tion from traffic on morbidity is also well documentedwith a variety of negative respiratory [3], cardiovascular [4]and reproductive effects [5] and lung cancer [6]. A recentcomprehensive review concluded that there is sufficientevidence to infer a causal role for TRAP in the exacerba-tion of asthma in children and suggestive evidence of itsrole in the onset of asthma in children [7]. A number ofpollutants (e.g. CO, NOX, and PM components) that areroutinely measured at fixed regulatory monitoring siteshave been used to represent exposure to TRAP. However,regulatory monitoring data cannot capture the fine-scalespatial pollutant gradients associated with vehicle emis-sions. Most of the recent epidemiological studies assessingTRAP have used methods with higher spatial resolutionto provide individual-level exposure estimates. Thesemethods generally estimate different surrogates of the traf-fic mixture (e.g. NO2, Black Carbon) derived from disper-sion or land use regression (LUR) models [8]. Despitethese advances in TRAP exposure assessment, none of the* Correspondence: hind.sbihi@ubc.ca1School of Population and Public Health, University of British Columbia, 2206East Mall, Vancouver, BC, Canada V6T 1Z3Full list of author information is available at the end of the article© 2013 Sbihi et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Sbihi et al. Environmental Health 2013, 12:48http://www.ehjournal.net/content/12/1/48pollution?An analysis of determinants of hopanessurrogate pollutants measured or modeled are specific tovehicle emissions.In addition to the lack of specificity, these methodscharacterize ambient levels and do not consider indoor in-filtration. Since individuals in North America spend anaverage of 87% of their time indoors [9,10] and many pol-lutants readily penetrate indoors, a significant proportionof total exposure to outdoor-generated pollutants occursindoors. Quantifying the PM infiltration efficiency (Finf ) inresidences can help characterize indoor concentrationsand reduce exposure misclassification [11] since Finf canvary 2 to 10- fold between houses that have the same am-bient concentrations [11-13].Unfortunately, methods for estimating Finf in residencesrequire home-specific indoor and outdoor sampling,which makes estimating Finf in large epidemiological stud-ies virtually impossible. To overcome this limitation, pre-diction models of Finf have been developed [12,13]. Whilethese models have shown promise, they have generallybeen developed for individual cities using relatively smallsample sizes and therefore may not be transferable toother locations.Hence, current approaches to estimate individual TRAPexposures (LUR, dispersion model, geostatistical methods)have two consistent limitations: (i) TRAP surrogates arebased on non-specific pollutant measures; (ii) modeled es-timates predict concentrations outside the home whilemost exposure occurs indoors.Settled house dust is a sink and repository for particle-bound material and semi-volatile organic compounds.Despite the variations that occur in sampling, dust mea-sures have formed the backbone of epidemiological studiesof multiple biological agents [14]. Indeed, house dust pre-sents the advantage of providing one matrix for the evalu-ation of multiple agents which is a reasonable proxy fortime-integrated exposure [15]. While the accumulation ofhouse dust depends on several factors (e.g. infiltration effi-ciency, indoor and outdoor pollutant sources, cleaningpractices, sampling surface), dust concentrations and load-ings of pollutants show less variation over time than do in-door air concentrations, therefore, dust sampling is aparticularly useful tool in studies of chronic exposures[16]. Measurement of airborne pollutants, for example ofhopanes in PM2.5, is typically only conducted for shorttime intervals, use of air samples to assess chronic expo-sures would require longer sampling intervals or repeatedmeasurements, features that are typically limited due tologistical (participant burden) or financial constraints.Using house dust as a marker for indoor inhalable hazardsand infiltrated pollutants of outdoor origin (e.g. polycyclicaromatic hydrocarbons (PAHs) from vehicle exhaust)would represent a useful and readily available exposure as-sessment tool. A good tracer of TRAP in dust would be achemical: 1) for which the major source is vehicleemissions; 2) for which emissions are correlated withother motor vehicles constituents; 3) that can be measuredat low levels for reasonable cost; and 4) that can be mea-sured with accuracy and stability.One such group of tracers may be the hopanes, a classof organic compounds with 27 to 35 carbon atoms in anaphthenic structure [17]. Hopanes are not found ingasoline and diesel fuel because they are in the higherboiling fraction of petroleum, but are present in engineoil lubricants [18]. Hopanes are tracers of primary ve-hicular exhaust aerosols in ambient air [19], particularlyon account of their relative stability and non-volatile na-ture in the atmosphere [20]. Schauer et al. showed thathopanes and steranes could be used to distinguish dieseland gasoline engine emissions from other combustionsources [21]. These relatively stable species can serve asunique tracers to determine the contribution of dieseland gasoline vehicles to particulate matter concentra-tions measured in outdoor air [3,22]. Measurement ofhopanes in settled house dust may therefore be useful toestimate time-integrated exposure to TRAP, while alsoaccounting for variability in infiltration. Our overallgoals were to evaluate the potential utility of hopanes asTRAP exposure surrogates by determining (i) whetherthe hopane mixture in house dust had similar compos-ition as that in outdoor air and (ii) the relationship be-tween hopanes in settled house dust with predictors ofTRAP spatial variability.MethodsWe utilized indoor dust measurements from fiveCanadian cities spanning four provinces (from West toEast: Vancouver (2.31 Million inhabitants), Edmonton(1.16 Million), Winnipeg (0.73 Million), Toronto (5.58Million) and Windsor (0.32 Million) [23]) in order toensure sufficient variability in hopane levels. Specific-ally, we conducted a city-level analysis where both city-specific and harmonized LUR variables across all citieswere examined. These analyses also included covariatesidentified in housing characteristics surveys that wereadministered in the different studies used for thisinvestigation.PopulationSamples were collected in three separate studies, brieflydescribed here, in which house dust was collected frominside homes of study participants:1 the Canadian Healthy Infant LongitudinalDevelopment (CHILD) study is a prospectivelongitudinal, birth-cohort study that has enrolled3650 families from Vancouver, Edmonton,Winnipeg, and Toronto between 2009 and 2012.Homes that (i) underwent home assessment whenSbihi et al. Environmental Health 2013, 12:48 Page 2 of 11http://www.ehjournal.net/content/12/1/48the child was at an age of 3-months; (ii) completed thequestionnaires on environmental factors, and (iii) haddust samples with sufficient dust mass for the analysisof several agents (endotoxins, β-glucans, and hopanes)were selected while ensuring balanced samplerepresentation from the four CHILD cities. Thus, 120homes analyzed for the suite of hopane monomers byDecember 2010 were included in this study;2 The Toronto Child Health Evaluation Questionnaire(TCHEQ) with 1,500 subjects from a nested case–control study were randomly selected from a largersurvey of 5,619 students who completed a screeningsurvey for respiratory disease [24]. Within thisnested study, a sub-sample of 50 homes wereinspected in 2006/2007 and underwentmeasurement of indoor/outdoor concentrations oftraffic related pollutants. From these, only 24 homes,included in this study, with sufficient dust mass forthe analysis of allergens (Der p, Der f, Ergosterol andGlucans) were also analyzed for hopanes. [25];3 During 2005/2006, Health Canada and theUniversity of Windsor conducted a personalexposure study in Windsor [26] (the WindsorOntario Exposure Assessment Study, WOEAS), inwhich 48 households were randomly recruited fromthe larger Windsor Children’s Respiratory HealthStudy [27] and where preference was given tospatially distributed households across Windsor.From these households, all homes with sufficienthouse dust mass were selected (n=27) to examinethe hopanes levels in house dust.HopanesDust samples: collection and analysisHouse dust samples were collected from the living roomsin all the homes included in the study. Participants wereasked not to vacuum during the week prior to the homevisit. Sampling was conducted by trained technicians whowere instructed to measure the sampling area, note thetype of surface and collect a pre-determined amountof dust.The WOEAS and TCHEQ sampling protocols were simi-lar as technicians vacuumed a 4 m2 section of floor for aperiod of 4 minutes or until at least one gram of dust wascollected and used high volume devices. In WOEAS, settleddust was collected using the High Volume Surface Sam-pling System (HVS3) vacuum [26], while TCHEQ used theShop-Vac vacuum (Model: QAM70, 7.0 Amps), anotherhigh volume device, equipped with Dust Sampling Socks(X-Cell 100, Midwest Filtration, Cincinnati, OH, USA). InCHILD, house dust samples were collected using a stan-dardized consumer model vacuum cleaner (Sanitaire,model S3686) fitted with a dust collection device designedespecially for the CHILD study with the goal of increasingthe collection efficiency without having to vacuum the en-tire area. This modified collector included slots for twonylon filter thimbles, thereby doubling the filtration surfaceand was constructed from machined aluminum, and outfit-ted with Teflon wheels to prevent marring of non-carpetedflooring, and to maintain the collection slot at a fixed dis-tance from such floors. The sample was taken from a 2 m2area by making seven passes of the nozzle over adjacentswaths of flooring. Only hopane concentrations in the fam-ily room were considered in the analysis since homes fromWOEAS and TCHEQ studies did not provide samplesfrom the bedroom and the bedroom samples in CHILD in-cluded a mixture of floor and bed samples.All dust samples were sieved into <150 μm size fractionsand reweighed for analysis. The sieved fractions werealiquoted and frozen at −80°C pending further analysis atthe Environment Canada laboratory operated the AirQuality Research Division in Downsview, Ontario. Extrac-tion in an isooctane solution was conducted with an ASE200 (Accelerated Solvent Extractor) followed by solventreduction using a Zymark TurboVap. Recovery standardswere added to the dust/solvent matrix before extractionand blow down. A suite of organic compounds were quan-tified by tandem Gas chromatography–mass spectrometry,including eleven hopane monomers. The final dust-relatedmetric for each of the individual hopanes and the sum ofall 11 monomers was expressed as the concentrations pergram of sieved dust (ng/mg), thereby correcting for differ-ences in the total amount of dust collected in each sample.Outdoor hopane measurementsThe composition of hopane mixtures, expressed as theabundance of the highest concentration monomer (17α(H),21β(H)-Hopane) relative to the sum of the concentrationsof all 11 measured monomers, was compared betweenavailable PM2.5 outdoor air samples in Vancouver, Edmon-ton, Toronto and Windsor with house dust samples for thesame cities. In all cities, one 24-hr ambient PM2.5 samplewas collected at Environment Canada national monitoringnetwork (NAPS) sites [28] within the same week in themonths of January, April, July and October 2010. The samesuite of hopane monomers available in the dust sampleswere quantified at the NAPS Environment Canada Labora-tory in Ottawa, Ontario, by thermal desorption gas chro-matography mass spectrometry [29] from punches ofarchived pre-fired quartz filters. Dust and air samples werematched by city and season.Geographic predictor variablesHarmonized geographic data were derived to allow forpooled analysis of all dust hopane measurements from allfive cities where samples were collected. We generated 30variables in 5 categories that are often used in develop-ment of LUR models for TRAP [8,30]. Subcategories wereSbihi et al. Environmental Health 2013, 12:48 Page 3 of 11http://www.ehjournal.net/content/12/1/48generated to characterize the street network, land use, andpopulation density within circular buffer sizes where theradius was set to represent close, medium and large geo-graphical areas around each home where the house dustsampling was conducted (Table 1). Highways and majorroads were defined by standard road classification categor-ies (DMTI Spatial Inc., Markham, Ontario), with categor-ies 1 (expressway), 2 (principal highway), and 3 (secondaryhighway) all considered highways (RD1), and category 4 asmajor roads (RD2). We also examined land use, elevationrelative to sea level and the distance to the nearest featureswithin the street network.All variables in each category were derived from a singlespatial dataset in vector format. Input files for the RoadLength and Land Use were taken from the 2006 DMTISpatial (Markham, Ontario) data files. Population Densitycategories were generated from the 2006 census distrib-uted by Statistics Canada and converted into point files atthe block level. Digital Elevation Data was obtained fromGeoBase in raster format at the municipal level. All inputfiles were manipulated in ArcGIS 10 (ESRI, Redlands, CA)to produce variable layers in raster format at 10 m reso-lution, except for the digital elevation model where thefinest available resolution was 30 m. From the latter data,relative elevation was defined as the mean centered city-specific elevation.We also extracted city-specific variables that had pre-viously been extracted and used in the development ofLUR models for NO2 in each of the cities [31-34] (seeAdditional file 1). Since these LUR variables had beenused to explain variability in outdoor NO2 in these citieswe therefore expected that they would explain variabilityin dust hopanes concentrations.QuestionnairesWe also included data from questionnaires delivered ineach of the indoor measurement studies on housing char-acteristics and lifestyle factors, which may be related to in-door hopane variability and/or infiltration (Table 2).For homes that were part of the CHILD study, home in-formation was gathered from both a questionnaire com-pleted by the parents and the home inspection conductedby research technicians. For homes that were part ofTCHEQ, a large amount of housing characteristics datawere also available from a questionnaire that was adminis-tered at study baseline (633 questions). Finally, from theWOEAS homes for which information on a wide range ofhousing characteristics and time-activity patterns was col-lected twice, we used the baseline questionnaire. Thequestionnaires included questions that were unique toeach cohort as well as other questions common across allstudies (see Table 2), which were recorded to generate aset of harmonized data. Harmonized variables includeddata related to the season (defined using heating degreeday) based on the date when samples were collected, thetype of floor (recoded as smooth for hard wood, vinyl andother smooth surfaces, carpets for rugs and carpets, andmixed for samples collected from both smooth and carpetsurfaces), the type of household (single or multifamily),the presence or absence of a garage, the type of garage(attached or detached), the presence of air conditioning(central or in a wall or portable unit), the frequency of useTable 1 Harmonized GIS dataCategory(Number of variables)Description Sub-category Buffer radii (m) Source/typeRoad Length Total length of two road types RD1 (Highways) 50, 100, 500, 1000 DMTI Road Network (Polyline)(8) RD2 (Major Roads)Land use Total area of different land use types (ha) COMM (commercial) 100, 500,1000 DMTI spatial data(12) OPEN (polygon)PARKINDUS (industrial)Distance to nearest feature Distance to nearest road type (m) Dist_RD1Dist_RD2(6)Distance to nearest land use type (m) Dist_Comm DMTI spatial data (polygon)Dist_OpenDist_ParkDist_IndusPopulation density Density of the population (persons/hectare) POPDENS 100, 1000, 2500 Block level census data (point file)(3)Geographic position Elevation (m) ELEV Geobase DEM (raster)Sbihi et al. Environmental Health 2013, 12:48 Page 4 of 11http://www.ehjournal.net/content/12/1/48of air conditioning (recoded as never, sometimes and fre-quently), the cleaning frequency (recoded as rarely, some-times and frequently), and the usage of windows (recodedand grouped from different questions in the CHILD ques-tionnaire) coded into 5 categories: usually open/sheer;covered with blinds or curtains; sealed; open daytime/cov-ered nighttime; other.Statistical analysisWe first analyzed the association between the mixture ofhopanes in outdoor air and indoor dust by comparing therelative abundance of the most abundant monomer. Spe-cifically, we calculated the ratio of the concentration of17α(H), 21β(H)-Hopane to the total concentration of theeleven monomers. We then compared this relative abun-dance between the outdoor air and indoor dust samples ineach city, and examined this association after accountingfor temperature and evaluating multicollinearity betweenpredictors (assessed by the variance inflation factor) in lin-ear regression modelsAfter examining the distribution of hopane concentra-tions in a pooled analysis of dust samples from all cities(hereafter “pooled analysis”) and separately within eachcity (“city-specific analysis”), we applied a log transform-ation to the total hopane concentration (i.e. sum of the 11monomers) distribution across all cities and within eachcity. Prior to examining the association between totalhopane concentrations and GIS variables in the pooledanalysis, we fit a random effects model with a randomintercept at the city level to assess the between- andwithin-city variability. Both in the pooled and city-specificanalyses, questions on lifestyle factors and housing charac-teristics were examined in bivariate analysis as potentialconfounders or effect modifiers for the hopane – geo-graphic predictor relationships.The same model building approach described byHenderson et al. [33] to generate physically meaningfulpredictive models was adopted and consisted of the fol-lowing steps: (1) Rank all variables by the absolutestrength of their correlation with the hopane concentra-tion; (2) Within each sub-category (e.g. all buffer sizes forhighway lengths), keep only the highest-ranking; (3) toavoid collinearity examine the correlations between allGIS predictors retained from step 2 as well as question-naire variables using 0.6 as a cut-off value ; (4) enter allremaining variables into a stepwise linear regression; (5)remove from the available pool any variables that have in-significant t-statistics and variables that show a directionof effect opposite of a priori hypotheses. These five stepsfollow previous LUR models [8] and the general ap-proaches often used in determinants of exposure model-ing, for example in assessment of occupational exposures[35]. Steps 4 and 5 were repeated until a parsimoniousfinal model that best explained the variations in indoordust hopanes levels was obtained.The study methodology was reviewed and approved byboth the University of British Columbia BehavioralResearch Ethics Board (ethics certificate no-H11-03231)and the Clinical Research Ethics Board (H07-03120).Table 2 Descriptive summary of questions found(as shown with a check mark) in the questionnairesdelivered during home visits, recoded for analysis in thepooled investigation of hopanes in dust and land usedeterminants of traffic pollutionQuestion type CHILD TCHEQ WOEASEmissions sources within 100 mof the home√× ×Factory 2%Gas station 11.3%Parking 15.6%Construction site 23.5%Shoe removal √× ×Yes 94%No 6%Type of floor √ √ √mixed 4% 83% 4%smooth 21% 17% 36%carpets 75% 60%Cleaning frequency √ √ √Rarely 14% 4% 56%Moderately 80% 71%44%Frequently 6% 25%Window usage/type √ √ √Usually open/sheer 15% 37.5% 33.3%Covered with blinds/curtains 42.5% 14.8% 63%Sealed 34% 14.8% 3.7%Opened daytime/ closed night Other 2% 14.8% 0%Garages √ √ √Yes 46% 17% 52%No 54% 83% 48%Type of house √ √ √single 64%100% 100%multifamily 36%Air conditioning √ √ √Yes 40%100%81%No 60% 11%Frequency of AC use √ √ √Frequently 21% 42% 15%Sometimes 19.5% 46% 7%Don’t know 59.5% 12%Never 0 78%Sbihi et al. Environmental Health 2013, 12:48 Page 5 of 11http://www.ehjournal.net/content/12/1/48ResultsOutdoor vs. indoor hopane concentrations comparisonAll samples were above the GC/MS limit of detection(LOD). In all cities the monomer 17α(H), 21β(H)-Hopanewas consistently detected and showed the highest abun-dance in the suite of analyzed compounds, therefore thecomparison of outdoor and indoor hopane ratios wasperformed using this monomer relative to the sum of allmonomers. The sampling from the air monitoring stationswas conducted in 2010 at fixed time points which result inconcentrations from air samples with a discrete distribu-tion (Figure 1) compared with the house dust sampleswhich were collected throughout the year. In air, the rangeof the 17α(H), 21β(H)-Hopane relative abundance (0.2 to0.4) generally corresponded to the same relative abun-dance in house dust. The correlation of the 17α(H),21β(H)-Hopane relative abundance in outdoor air andhouse dust was moderately strong, yet significant (r=0.48,p<0.05).After excluding an outlier sample (see Figure 1 datapoint near zero where the ratio in 17α(H), 21β(H)-Hopanewas depleted due to a very low concentration in all mono-mers), we also examined the relation between the outdoorand the indoor relative abundance in linear regression ac-counting for the effect of season, and found a strongerstatistically significant relationship (slope=0.92, t=6.1)compared to the association without adjustment for sea-son (slope=0.72, t=5.9). The correlation was still signifi-cant when the outlier was included. The effect of seasonwas stronger during the spring and summer (r>0.5) thanduring the fall and winter (r<0.5).Pooled and city-specific resultsHopane levels in individual homes varied from a low of0.4 ng/mg of dust in a Toronto home to a high of 41.8ng/mg of dust in a Vancouver sample, after excluding anoutlier in Toronto from the CHILD study where theconcentration was 160.3 ng/mg (more than 29 timeshigher than the median); for this home we examined theland use characteristics, the road network, and potentialoutdoor sources as indicated in the questionnaire anddid not find any difference that would explain such ahigh concentration. Analyses were thus run with andwithout this sample (Table 3).Windsor had the lowest overall indoor hopane levels witha mean level of 5.8 ng/mg (GM= 5.1 ng/mg, GSD =1.8)while the sample of homes in Vancouver showed thehighest mean concentration of 9.3 ng/mg (GM= 6 ng/mg,GSD= 2.9) when the high (Toronto) outlier was excluded(if the Toronto outlier was retained, then Toronto wasranked first with an AM=9.9 ng/mg).Pooled analysisAfter fitting a null random effect model, the intraclass cor-relation of 0.007 indicated that city clustering would notcontribute to explaining the variability in total hopaneconcentrations. We therefore built a model without usinga city-specific random intercept in the regression analysisand all samples were treated as independent.From the harmonized questions, only relative elevationand heating degree days at the time of dust collectionshowed a statistically significant relationship with hopaneconcentrations. Distance to highway (DIST_RD1) had aFigure 1 Association between outdoor air and house dust hopane major monomer (17α(H), 21β(H)-Hopane) relative abundance.Sbihi et al. Environmental Health 2013, 12:48 Page 6 of 11http://www.ehjournal.net/content/12/1/48statistically significant association with hopane concentra-tions, but its direction of effect was opposite to a priori ex-pectations and was therefore excluded from the model.The final model with relative elevation and heating degreedays as predictors explained only 6% of the total variabilityin the hopanes concentration. Including distance to high-way did not appreciably improve the amount of explainedvariability (adjusted R2 =0.08). Excluding the high outlierhome in Toronto led to a model with the addition of thepresence of an AC unit in the home along with the samepredictors as above, but with less overall variabilityexplained (adjusted R2= 0.04).City-Specific modeling resultsGiven the availability of LUR models for predicting NO2in each study area, we extracted the NO2 concentrationat the geocoded participants’ home addresses and exam-ined the correlation of hopane concentrations in housedust with city-specific LUR NO2 estimates in each city.Results (Table 4) indicated no statistically significant as-sociations except in Windsor (r=0.44, p<0.05) andEdmonton (r=0.58, p<0.05).Leveraging the availability of city-specific LUR models,we further examined separately for each city the associ-ation between hopane concentrations and the variablesthat were used both in the city-specific LUR models de-scribing the NO2 levels (see Additional file 1) and thosethat we generated for the pooled analysis (Table 2). Theamount of variability explained in each city varied from10% in Vancouver to 80% in Edmonton (Table 3).Overall, in each city the determinants of indoor dusthopanes were predominantly related to home-specific fac-tors (cleaning, use of AC, shoe removal) and meteorology,except for Windsor where the final model included thelength of major roads in a 100m buffer (Table 3). InToronto, the spatial variability provided by the TCHEQTable 3 City-specific determinants of hopane concentrations in house settled dustCity Final model with regression coefficients Partial R2 Model Adj.R2Edmonton log (hopanes) = 3 – 0.13 cleaning frequency 0.78 0.80- 1.5 Smooth Flooring 0.78−0.15 Air Conditioning 0.35Toronto Log (hopanes) = 2.69 -1.03 Smooth Flooring 0.29 0.45−0.008 Elevation 0.13+ 0.88 Attached Garage −0.66 Detached Garage 0.13%*Windsor** log (hopanes) = 5.6 + 0.5 elevation 0.36 0.39+ 0.17 RD1_100 0.13Winnipeg log (hopanes) = 1.45 – 0.057Heating degree days 0.17 0.33– 1.33 multifamily house 0.16Vancouver log (hopanes) = 1.9 – 0.09 Heating Degree Days 0.09 0.10– 0.07 Shoe removal 0.07* The Garage variable has three categories: No garage, Attached garage and Detached garage.** In Windsor, elevation and distance to the Ambassador Bridge were strongly and significantly correlated. An alternative model with Distance to AmbassadorBridge yielded similar results, yet with smaller R2Table 4 Geometric Mean (GM) and Geometric Standard Deviation (GSD) of total Hopanes concentrations in the studypopulation, by room, by home and correlation a with city-specific modeled NO2CityHomes Hopanes concentration (ng/mg) Pearson correlationN(number of homes)Bedroom Family room Average Between NO2 and family roomn GM (GSD) n GM (GSM) N GM (GSD) r (p-value)Winnipeg (CHILD) 26 23 4.9 (2.1) 21 5.8 (2.1) 40 5.3 (2.3) 0.04 (n.s.)Edmonton (CHILD) 15 12 4.7 (2.7) 14 4.1 (2.0) 26 4.5 (2.3) 0.58 (0.03)Vancouver (CHILD) 65 56 7.4 (2.2) 54 6 (2.9) 90 6.7 (2.6) -0.12 (n.s.)Toronto (CHILD) 14 13 5.9 (1.9) 12 7.7 (2.9) 22 6.6(2.3) 0.02 (n.s.)Windsor (WOAES) 27 NA 27 5.1 (1.8) NA 0.44(0.02)Toronto (TCHEQ) 24 NA 24 4 (2.5) NA 0.18 (n.s.)Abbreviations: n total number of samples, n.s. not statistically significant association.Sbihi et al. Environmental Health 2013, 12:48 Page 7 of 11http://www.ehjournal.net/content/12/1/48samples was very limited as all homes were within a re-stricted geographic area within the city. Hence, an add-itional sub-analysis was run for Toronto with only theCHILD homes included. This model (not shown) did re-tain GIS variables (open area within 1000 m buffer andelevation) as well as variables related to other possiblesources of hopane emissions (garage type, presence of aconstruction site within 100 m) and finally home-specificfactors (i.e. the type of floor surface) and explained 86% ofthe overall variability in indoor dust hopanes. After ex-cluding the house with the outlier concentration value,however, the final model, with an R2 = 0.3, had exactly thesame predictors as those shown in Table 3 where samplesfrom both the TCHEQ and CHILD study homes inToronto were included.While the association of hopanes indoors in relation toGIS variables typically used as surrogates for TRAP wasonly modeled for samples collected in living rooms, Table 4shows the concentration in each city by room type andthe number of homes (from the CHILD study) where tworooms were sampled. In CHILD, participating householdsprovided dust samples from the living room as well as asecond composite sample from subject child’s mattressand adjacent flooring. The ranking by decile showed thatthe hopane concentration in the living rooms was signifi-cantly greater than that found in the bedrooms.DiscussionAssessing indoor levels of TRAP through the collectionand analysis of settled house dust is a new area of studyand has the potential to reduce the misclassification andincrease the specificity of exposure. In this investigation,we compared hopanes in dust and ambient air and withGIS-derived land use variables. This is the first investiga-tion of hopanes collected in house settled dust. Theavailability of contemporaneous cohort studies (CHILD,TCHEQ and WOEAS) offered a unique opportunity togather a sample of 171 homes where dust was collectedusing similar protocols in 151 living rooms and wherehopanes were analyzed by GC/MS at the same labora-tory using a standardized protocol. Samples were col-lected from different settings ranging from highly urbanlocations such as Toronto to smaller and less denselypopulated cities such as Winnipeg, while also includingmajor transit hubs such as Windsor, the site of a majorCanadian-American truck border. Furthermore, all thecities had previously developed LUR models which rea-sonably predicted traffic related NO2 spatial variability(from 66% in Vancouver to 81% in Edmonton [31,33].Still, these homes represent only a small fraction of thetotal homes in each city and even of the homes includedin each of the studies. Further, different numbers ofhomes were included in the different cities. We aretherefore unable to make conclusions regarding the rep-resentativeness of the measured hopanes levels and in-stead focused on the variability within and betweencities and the extent to which this variability could beexplained by various potential determinants.We demonstrated that hopanes can be consistentlydetected in house dust samples regardless of the type ofcity and the dust collection location. In addition, after con-trolling for heating degree days and its impact on infiltra-tion, the major hopane monomer relative abundance inhouse dust and outdoor air samples were significantly cor-related (r = 0.48), suggesting similar hopane sources in thetwo samples, but there remains substantial unexplainedvariability in indoor levels. This comparison had relativelygood external validity given that the ambient monitoringsites were located to capture urban background concen-trations rather than hot spots and since samples were col-lected in and matched for all seasons. This correlationwas stronger in the summer compared to the winter,suggesting an impact of infiltration as windows are morelikely to be opened on warmer days. Since hopanes inhouse dust accumulate over relatively longer periods oftime compared with hopanes in air samples and may haveundergone many changes and cycles in temperature, it islikely that the seasonal effect shown in the literature[36-38] may not hold in this context. In addition, dustsampling, which often is a readily available matrix for sam-pling multiple agents in epidemiological studies, includinghopanes as demonstrated in this study, does not representsimilar constraints (e.g. logistics) as those imposed by par-ticle infiltration measurements.Examining associations between hopane concentra-tions and geographic predictors in a pooled analysis in-dicated that only a small degree of variability in hopaneconcentrations in dust was explained by the final model.Further, in this analysis, higher levels of road variableswere linked to lower levels of hopanes. Despite the ad-vantages of pooling data from different cohorts, this ef-fort was hindered by the absence of consistency in thesupplementary data collected via questionnaires sinceeach study used its own set of questions. While weinspected each question and the research technicians’notes for each sample of house dust collected in orderto generate harmonized variables that could affect thehopane concentration in house dust, recoding variablesmay have resulted in a loss of specificity.Unlike the pooled analysis, the city-specific analysisprovided more insight into the utility of hopanes as pos-sible markers for TRAP as a moderate to large amountof variability in the total hopane concentration in housedust was explained in each model. This analysis, how-ever, was hindered by the lack of consistency betweencities in terms of main predictors of indoor hopaneconcentrations.Sbihi et al. Environmental Health 2013, 12:48 Page 8 of 11http://www.ehjournal.net/content/12/1/48We examined potential modifiers that could alter therelationship between LUR variables and hopanes in dustfor each city separately. In addition to geographic surro-gates of TRAP, all cities had at least one predictor ofhopane concentration related to the indoor environmentor home construction. Thus, the variability in settleddust hopane concentrations appears to be a function ofa mix of parameters that are not exclusively related totraffic emissions.For example, indoor hopanes in settled dust may also re-sult from coarse PM being tracked indoors. A recent ana-lysis of indoor PAHs indicated the potential importance ofthis pathway even after adjusting for carpeting, frequencyof vacuuming and indoor burning [39]. We examined theassociation in the city-specific analysis for all CHILD par-ticipating homes between shoe removal habits and hopaneconcentrations. We found that only Vancouver sampleswere correlated with shoe removal habits in the expecteddirection. Collection of supplementary field data remainsa crucial component for assessing the utility of hopanes inhouse dust since tracked dust seems to contribute tohopanes concentration in house dust. This informationwas only available in the CHILD homes, and could there-fore not be assessed in the pooled analysis.In our investigation we made a critical assumptionthat hopanes have few sources beyond engine oil lubri-cants as we were not able to find information on indoorhopane sources in the literature. Since hopanes arewidespread in recent and ancient sediments, they areconstituents of all mineral oil or petroleum-based lubri-cants and it is therefore possible that unaccounted forindoor sources were present.House dust remains an attractive metric for exposure as-sessment because it offers a matrix for multiple indoor con-taminants, both biological and chemical and both indoorand outdoor in origin, and can be stored for long time pe-riods, thus providing the opportunity to examine additionalresearch questions when necessary. The utility of hopanesin house dust as an indicator of infiltrated TRAP is limitedin the absence of better understanding of its deposition andstability in house dust. House dust is heterogeneous matrixwith a complex history in each home as it accumulates con-tributions from multiple sources including not only freshemissions of combustion-related particles but also roaddust which also contains hopanes. The mode of accumula-tion also contributes to the variability of vacuum dust. Sev-eral factors that may vary among study participants canaffect the concentrations of hopanes: cleaning practices andsampling surfaces (carpeted vs. non-carpeted) play a role inthe amount of chemicals that deposit inside the homes asshown in the city-specific analysis. In addition, the metricof exposure for hopanes still lacks consensus as hopanescan be measured in terms of loading (concentration nor-malized by surface area sampled) or expressed as the moretraditional approach of normalized concentration to massof dust collected. Differences in the choice of metric wouldrelate mostly to cleaning practices, which we have tried toaccount for in our investigation. Future investigations ofother species, such as PAHs, on their own or in combin-ation with hopanes, may offer additional insight into theutility of settled house dust as a surrogate for TRAPexposure.In our study, we compiled the information aboutpresence and frequency of use of air conditioning asthis has been shown to be an important predictor ofPM infiltration [40], but we found limited explanatorypower in both pooled and city-specific analysis. PM in-filtration varies with particle size, with a maximuminfiltration efficiency for diameters of approximately0.2-0.3 μm [41], while the size distribution of hopanesranges between 0.7 and 3.3 μm [42] which would implythat hopane infiltration efficiency may be low andmight therefore explain variability in the outdoor/in-door correlation [37]. We could expect that in pres-ence of higher concentrations of hopanes in ambientair (i.e. better ability to detect hopane monomers), theanalysis of relative abundance in ambient and indoorhopane would have shown less unexplained variability.ConclusionsOur results indicate that indoor dust hopane concentra-tions depend on both outdoor TRAP and on a variety ofhome-specific variables such as cleaning, floor type, andpresence of AC. This conclusion is supported by ouranalysis of the relative variation explained by LUR NO2compared to home-specific factors as we also found thatin some cities a correlation between hopanes and LURNO2 is only revealed when accounting for variation dueto such home-specific factors.We examined the utility of measurements of hopanesin house dust as exposure indicators for infiltrated,time-integrated, traffic-related pollutants. When com-bined with behavioral factors retrieved from question-naires, and geographic determinants, hopanes in housedust may have the potential to be used as surrogates forinfiltrated TRAP. Further characterization of the deter-minants of hopanes in house dust may result in an im-proved exposure measure for epidemiologic studies tomore precisely analyze relationships between TRAP andchronic health effects.ConsentWritten informed consent was obtained from the studyparticipant’s guardian/parent for the use of personal infor-mation kept confidential and only used for scientificobjectives.Sbihi et al. Environmental Health 2013, 12:48 Page 9 of 11http://www.ehjournal.net/content/12/1/48Additional fileAdditional file 1: City-specific GIS variables and buffer sizesextracted from LUR model surfaces.AbbreviationsCHILD: Canadian Healthy Infant Longitudinal Development; GIS: GeographicInformation Systems; GM: Geometric Mean; GSD: Geometric StandardDeviation; LUR: Land Use Regressions; TCHEQ: Toronto Child HealthEvaluation Questionnaire; TRAP: Traffic-Related Air Pollution; WOEAS: WindsorOntario Exposure Assessment Study.Competing interestsAll authors declare no competing financial interests.Authors’ contributionsHS formulated the research question, gathered spatial data and extractedquestionnaire data, conducted the statistical analysis and led the writing ofthe manuscriptMB guided the study design, provided spatial data forVancouver, helped formulate the study questions, gave critical input in dataconditioning for the analysis and revised the manuscript for its intellectualcontent. RA provided the spatial data for Edmonton and Winnipeg andguided the data harmonizing across different spatial databases. JB gatheredthe ambient air samples from the national air monitoring stations andoversaw the chemical analysis of hopanes samples in Environmental CanadaLaboratory. JC performed a preliminary analysis examining personal hopanesin dust and air samples. JB, JS, TK, RA, and MB are part of the ExposureWorking Group for the CHILD study where all questionnaires used in thisstudy were designed. This group, along with the site leaders and CHILD PIoversaw the collection and initial processing of CHILD dust samples. ST, PM,PS are site leaders for the CHILD study, they are responsible for thecoordination and training of research technicians and data collection. MS isthe Principal Investigator of the CHILD study and has revised the manuscriptdrafts. SD is the Principal Investigator of the TCHEQ study and revised themanuscript. AW has led the WOEAS, conceived of this study and participatedin preparing the spatial data for Windsor, she has given critical input to themanuscript drafts. All authors have contributed to this manuscript and givenapproval to the final version.AcknowledgementsThe authors would like to extend their gratitude to the following individualswho helped conduct this study:Xiaohong Xu's group at University of Windsor collected the Windsorsamples. Greg Evan's group at University of Toronto stored the TCHEQsamples. Zhimei Jiang and Luyi Ding at Environment Canada performed theGC/MS analysis of the dust and ambient filter data, respectively. DianaLefebvre and Justina Greene performed database extraction for CHILDquestionnaire data. Thank you to all participating families.The CHILD Study is funded by the Canadian Institutes of Health Researchand the Allergy, Genes and Environment (AllerGen) Network of Centres ofExcellence. WOAES and T-CHEQ studies were funded by Health Canada.H. Sbihi was funded by CIHR Banting and Best doctoral award and IzaakWalton Killam Memorial Pre-Doctoral Fellowship. S.E. Turvey was supportedby a Clinical Research Scholar Award from the Michael Smith Foundation forHealth Research and the Aubrey J. Tingle Professorship in PediatricImmunology.Author details1School of Population and Public Health, University of British Columbia, 2206East Mall, Vancouver, BC, Canada V6T 1Z3. 2Air Quality Research Division,Environment Canada, 4905 Dufferin Street, Toronto, Ontario, Canada M3H5T4. 3Faculty of Health Sciences, Simon Fraser University, 8888 UniversityDrive, Burnaby, BC, Canada V5A 1S6. 4Division of Respiratory Medicine, TheHospital for Sick Children, 555 University Avenue, Toronto, Ontario, CanadaM5G 1X8. 5Department of Pediatrics, Faculty of Medicine and Dentistry,University of Alberta, WC Mackenzie Health Sciences Centre, Edmonton,Alberta T6G 2R7, Canada. 6Dalla Lana School of Public Health, University ofToronto, 155 College St, Toronto ON M5T 3M7, Canada. 7Department ofMedicine, Faculty of Health Sciences, McMaster University, 1280 Main St W,Hamilton ON L8S 4K1, Canada. 8BC Children’s Hospital and Child & FamilyResearch Institute, 950 West 28th Ave, Vancouver, BC, Canada V5Z 4H4. 9AirHealth Science Division, Health Canada, 269 Laurier Avenue West, Ottawa,Ontario, Canada K1A 0K9.Received: 25 February 2013 Accepted: 12 June 2013Published: 19 June 2013References1. Jerrett M, Finkelstein MM, Brook JR, Arain MA, Kanaroglou P, Stieb DM,Gilbert NL, Verma D, Finkelstein N, Chapman KR, Sears MR: A Cohort Studyof Traffic-related Air Pollution and Mortality in Toronto, Canada.Environ Health Perspect 2009, 117:772–777.2. Künzli N, Kaiser R, Medina S, Studnicka M, Chanel O, Filliger P, Herry M,Horak F, Puybonnieux-Texier V, Quénel P, Schneider J, Seethaler R, VergnaudJ-C, Sommer H: Public-health impact of outdoor and traffic-related airpollution: a European assessment. Lancet 2000, 356:795–801.3. Brauer M, Hoek G, Van Vliet P, Meliefste K, Fischer PH, Wijga A, Koopman LP,Neijens HJ, Gerritsen J, Kerkhof M, Heinrich J, Bellander T, Brunekreef B: Airpollution from traffic and the development of respiratory infections andasthmatic and allergic symptoms in children. Am J Respir Crit Care Med2002, 166:1092–1098.4. Gan WQ, Tamburic L, Davies HW, Demers PA, Koehoorn M, Brauer M:Changes in residential proximity to road traffic and the risk of deathfrom coronary heart disease. Epidemiology 2010, 21:642–649.5. Brauer M, Lencar C, Tamburic L, Koehoorn M, Demers P, Karr C: A CohortStudy of Traffic-Related Air Pollution Impacts on Birth Outcomes.Environ Health Perspect 2008, 116:680–686.6. Beelen R, Hoek G, van den Brandt PA, Goldbohm RA, Fischer P, Schouten LJ,Armstrong B, Brunekreef B: Long-Term Exposure to Traffic-Related AirPollution and Lung Cancer Risk. Epidemiology 2008, 19:702–710.7. HEI Panel on the Health Effects of Traffic-Related Air Pollution: Traffic RelatedAir Pollution: A Critical Review of the Literature on Emissions, Exposure, andHealth Effects. HEI Special Report 17. Boston, MA: Health Effects Institute; 010.8. Hoek G, Beelen R, de Hoogh K, Vienneau D, Gulliver J, Fischer P, Briggs D: Areview of land-use regression models to assess spatial variation ofoutdoor air pollution. Atmos Environ 2008, 42:7561–7578.9. Klepeis NE, Nelson WC, Ott WR, Robinson JP, Tsang AM, Switzer P, Behar JV,Hern SC, Engelmann WH: The National Human Activity Pattern Survey(NHAPS): a resource for assessing exposure to environmental pollutants.J Expo Anal Environ Epidemiol 2001, 11:231–252.10. Leech JA, Nelson WC, Burnett RT, Aaron S, Raizenne ME: It’s about time: Acomparison of Canadian and American time-activity patterns[dagger].J Expo Anal Environ Epidemiol 2002, 12:427–432.11. Allen R, Wallace L, Larson T, Sheppard L, Liu L-JS: Evaluation of therecursive model approach for estimating particulate matter infiltrationefficiencies using continuous light scattering data. J Expos Sci EnvironEpidemiol 2006, 17:468–477.12. Clark NA, Allen RW, Hystad P, Wallace L, Dell SD, Foty R, Dabek-ZlotorzynskaE, Evans G, Wheeler AJ: Exploring Variation and Predictors of ResidentialFine Particulate Matter Infiltration. IJERPH 2010, 7:3211–3224.13. Hystad PU, Setton EM, Allen RW, Keller PC, Brauer M: Modeling residentialfine particulate matter infiltration for exposure assessment. J Expo SciEnviron Epidemiol 2009, 19:570–579.14. Arrandale VH, Brauer M, Brook JR, Brunekreef B, Gold DR, London SJ, MillerJD, Özkaynak H, Ries NM, Sears MR: Silverman FS. Takaro TK: ExposureAssessment in Cohort Studies of Childhood Asthma. Environ HealthPerspect; 2010.15. Lioy PJ, Freeman NCG, Millette JR: Dust: a metric for use in residential andbuilding exposure assessment and source characterization. Environ HealthPerspect 2002, 110:969–983.16. Egeghy PP, Quackenboss JJ, Catlin S, Ryan PB: Determinants of temporalvariability in NHEXAS-Maryland environmental concentrations,exposures, and biomarkers. J Expo Anal Environ Epidemiol 2004,15:388–397.17. Simoneit BRT: Application of Molecular Marker Analysis to VehicularExhaust for Source Reconciliations. Int J Environ Anal Chem 1985, 22:203.18. Pakbin P, Ning Z, Schauer JJ, Sioutas C: Characterization of Particle BoundOrganic Carbon from Diesel Vehicles Equipped with Advanced EmissionControl Technologies. Environ Sci Technol 2009, 43:4679–4686.19. Delfino RJ, Staimer N, Tjoa T, Arhami M, Polidori A, Gillen DL, Kleinman MT,Schauer JJ, Sioutas C: Association of Biomarkers of Systemic InflammationSbihi et al. Environmental Health 2013, 12:48 Page 10 of 11http://www.ehjournal.net/content/12/1/48with Organic Components and Source Tracers in Quasi-UltrafineParticles. Environ Health Perspect 2010, 118:756–762.20. Turlington JM, Olson DA, Stockburger L, McDow SR: Trueness, precision,and detectability for sampling and analysis of organic species inairborne particulate matter. Anal Bioanal Chem 2010, 397:2451–2463.21. Schauer JJ: Evaluation of elemental carbon as a marker for dieselparticulate matter. J Expo Anal Environ Epidemiol 2003, 13:443–453.22. Rogge WF, Hildemann LM, Mazurek MA, Cass GR, Simoneit BRT: Sources offine organic aerosol. 2. Noncatalyst and catalyst-equipped automobilesand heavy-duty diesel trucks. Environ Sci Technol 1993, 27:636–651.23. Population and Dwelling Count Highlight Tables, 2011 Census. http://www12.statcan.gc.ca/census-recensement/2011/dp-pd/hlt-fst/pd-pl/index-eng.cfm.24. Dell SD, Foty RG, Gilbert NL, Jerrett M, To T, Walter SD, Stieb DM: Asthmaand allergic disease prevalence in a diverse sample of Toronto schoolchildren: Results from the Toronto Child Health Evaluation Questionnaire(T-CHEQ) Study. Can Respir J 2010, 17:e1–e6.25. Loo CKJ, Foty RG, Wheeler AJ, Miller JD, Evans G, Stieb DM, Dell SD: Doquestions reflecting indoor air pollutant exposure from a questionnairepredict direct measure of exposure in owner-occupied houses? Int JEnviron Res Public Health 2010, 7:3270–3297.26. Wheeler AJ, Xu X, Kulka R, You H, Wallace L, Mallach G, Van Ryswyk K,MacNeill M, Kearney J, Rasmussen PE, Dabek-Zlotorzynska E, Wang D, PoonR, Williams R, Stocco C, Anastassopoulos A, Miller JD, Dales R, Brook JR:Windsor, Ontario exposure assessment study: design and methodsvalidation of personal, indoor, and outdoor air pollution monitoring. J AirWaste Manag Assoc 2011, 61:324–338.27. Dales R, Wheeler AJ, Mahmud M, Frescura A-M, Liu L: The Influence ofNeighborhood Roadways on Respiratory Symptoms Among ElementarySchoolchildren. J Occup Environ Med 2009, 51:654–660.28. Dabek-Zlotorzynska E, Dann TF, Kalyani Martinelango P, Celo V, Brook JR,Mathieu D, Ding L, Austin CC: Canadian National Air Pollution Surveillance(NAPS) PM2.5 speciation program: Methodology and PM2.5 chemicalcomposition for the years 2003–2008. Atmos Environ 2011, 45:673–686.29. Graham LA, Tong A, Poole G, Ding L, Ke F, Wang D, Caravaggio G, CharlandJ-P, MacDonald P, Hall A, Cheng Y, Brook JR: A comparison of directthermal desorption with solvent extraction for gas chromatography–mass spectrometry analysis of semivolatile organic compounds in dieselparticulate matter. Int J Environ Anal Chem 2010, 90:511–534.30. Dijkema MB, Gehring U, van Strien RT, van der Zee SC, Fischer P, Hoek G,Brunekreef B: A Comparison of Different Approaches to Estimate Small-Scale Spatial Variation in Outdoor NO2 Concentrations. Environ HealthPerspect 2011, 119:670–675.31. Allen RW, Amram O, Wheeler AJ, Brauer M: The transferability of NO andNO2 land use regression models between cities and pollutants.Atmos Environ 2011, 45:369–378.32. Jerrett M, Arain MA, Kanaroglou P, Beckerman B, Crouse D, Gilbert NL, BrookJR, Finkelstein N, Finkelstein MM: Modeling the intraurban variability ofambient traffic pollution in Toronto, Canada. J Toxicol Environ Health A2007, 70:200–212.33. Henderson SB, Beckerman B, Jerrett M, Brauer M: Application of land useregression to estimate long-term concentrations of traffic-relatednitrogen oxides and fine particulate matter. Environ Sci Technol 2007,41:2422–2428.34. Wheeler AJ, Smith-Doiron M, Xu X, Gilbert NL, Brook JR: Intra-urbanvariability of air pollution in Windsor, Ontario–Measurement andmodeling for human exposure assessment. Environ Res 2008, 106:7–16.35. Burstyn I, Teschke K: Studying the Determinants of Exposure: A Review ofMethods. Am Ind Hyg Assoc J 1999, 60:57–72.36. Olson DA, McDow SR: Near roadway concentrations of organic sourcemarkers. Atmos Environ 2009, 43:2862–2867.37. Olson DA, Turlington J, Duvall RM, McDow SR, Stevens CD, Williams R:Indoor and outdoor concentrations of organic and inorganic molecularmarkers: Source apportionment of PM2.5 using low-volume samples.Atmos Environ 2008, 42:1742–1751.38. Rutter AP, Snyder DC, Schauer JJ, Sheesley RJ, Olson MR, DeMinter J:Contributions of resuspended soil and road dust to organic carbon infine particulate matter in the Midwestern US. Atmos Environ 2011,45:514–518.39. Mahler BJ, Metre PCV, Wilson JT, Musgrove M, Burbank TL, Ennis TE, BasharaTJ: Coal-Tar-Based Parking Lot Sealcoat: An Unrecognized Source of PAHto Settled House Dust. Environ Sci Technol 2010, 44:894–900.40. Allen RW, Adar SD, Avol E, Cohen M, Curl CL, Larson T, Liu L-JS, Sheppard L,Kaufman JD: Modeling the Residential Infiltration of Outdoor PM2.5 inthe Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air).Environ Health Perspect 2012, 120(6):824–830.41. Sarnat SE, Coull BA, Ruiz PA, Koutrakis P, Suh HH: The Influences ofAmbient Particle Composition and Size on Particle Infiltration in LosAngeles, CA, Residences. 2006, 56:186–196.42. Huang X-F, Chen D-L, Lan Z-J, Feng N, He L-Y, Yu G-H, Luan S-J:Characterization of organic aerosol in fine particles in a mega-city ofSouth China: Molecular composition, seasonal variation, and sizedistribution. Atmospheric Research 2012, 114–115:28–37.doi:10.1186/1476-069X-12-48Cite this article as: Sbihi et al.: A new exposure metric for traffic-relatedair pollution? an analysis of determinants of hopanes in settled indoorhouse dust. Environmental Health 2013 12:48.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitSbihi et al. Environmental Health 2013, 12:48 Page 11 of 11http://www.ehjournal.net/content/12/1/48

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.52383.1-0223294/manifest

Comment

Related Items