UBC Faculty Research and Publications

Spatiotemporal air pollution exposure assessment for a Canadian population-based lung cancer case-control… Hystad, Perry; Demers, Paul A; Johnson, Kenneth C; Brook, Jeff; van Donkelaar, Aaron; Lamsal, Lok; Martin, Randall; Brauer, Michael Apr 4, 2012

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

Item Metadata

Download

Media
52383-12940_2011_Article_526.pdf [ 673.69kB ]
Metadata
JSON: 52383-1.0220554.json
JSON-LD: 52383-1.0220554-ld.json
RDF/XML (Pretty): 52383-1.0220554-rdf.xml
RDF/JSON: 52383-1.0220554-rdf.json
Turtle: 52383-1.0220554-turtle.txt
N-Triples: 52383-1.0220554-rdf-ntriples.txt
Original Record: 52383-1.0220554-source.json
Full Text
52383-1.0220554-fulltext.txt
Citation
52383-1.0220554.ris

Full Text

RESEARCH Open AccessSpatiotemporal air pollution exposure assessmentfor a Canadian population-based lung cancercase-control studyPerry Hystad1*, Paul A Demers2, Kenneth C Johnson3, Jeff Brook4, Aaron van Donkelaar5, Lok Lamsal6,Randall Martin7 and Michael Brauer8AbstractBackground: Few epidemiological studies of air pollution have used residential histories to develop long-termretrospective exposure estimates for multiple ambient air pollutants and vehicle and industrial emissions. Wepresent such an exposure assessment for a Canadian population-based lung cancer case-control study of 8353individuals using self-reported residential histories from 1975 to 1994. We also examine the implications ofdisregarding and/or improperly accounting for residential mobility in long-term exposure assessments.Methods: National spatial surfaces of ambient air pollution were compiled from recent satellite-based estimates(for PM2.5 and NO2) and a chemical transport model (for O3). The surfaces were adjusted with historical annual airpollution monitoring data, using either spatiotemporal interpolation or linear regression. Model evaluation wasconducted using an independent ten percent subset of monitoring data per year. Proximity to major roads,incorporating a temporal weighting factor based on Canadian mobile-source emission estimates, was used toestimate exposure to vehicle emissions. A comprehensive inventory of geocoded industries was used to estimateproximity to major and minor industrial emissions.Results: Calibration of the national PM2.5 surface using annual spatiotemporal interpolation predicted historicalPM2.5 measurement data best (R2 = 0.51), while linear regression incorporating the national surfaces, a time-trendand population density best predicted historical concentrations of NO2 (R2 = 0.38) and O3 (R2 = 0.56). Applying themodels to study participants residential histories between 1975 and 1994 resulted in mean PM2.5, NO2 and O3exposures of 11.3 μg/m3 (SD = 2.6), 17.7 ppb (4.1), and 26.4 ppb (3.4) respectively. On average, individuals livedwithin 300 m of a highway for 2.9 years (15% of exposure-years) and within 3 km of a major industrial emitter for6.4 years (32% of exposure-years). Approximately 50% of individuals were classified into a different PM2.5, NO2 andO3 exposure quintile when using study entry postal codes and spatial pollution surfaces, in comparison toexposures derived from residential histories and spatiotemporal air pollution models. Recall bias was also presentfor self-reported residential histories prior to 1975, with cases recalling older residences more often than controls.Conclusions: We demonstrate a flexible exposure assessment approach for estimating historical air pollutionconcentrations over large geographical areas and time-periods. In addition, we highlight the importance ofincluding residential histories in long-term exposure assessments.For submission to: Environmental HealthKeywords: Air pollution, Canada, Exposure assessment, Lung cancer, Residential mobility, Spatiotemporal* Correspondence: phystad@gmail.com1School of Population and Public Health, University of British Columbia, 2206East Mall, Vancouver, BC V6T 1Z3, CanadaFull list of author information is available at the end of the articleHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22© 2012 Hystad et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative CommonsAttribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction inany medium, provided the original work is properly cited.BackgroundExposure to ambient air pollution is a suspected risk fac-tor for lung cancer [1-6]. Due to the long latency periodsassociated with lung cancer, epidemiological analyses areparticularly challenging, especially for air pollution wherespatial and temporal variation in both residential mobilityand air pollution concentrations may produce significantexposure misclassification if not properly incorporatedinto the exposure assessment approach.Residential mobility data are required for accurate long-term air pollution exposure assessments, but due to thedifficulties in obtaining this information, residential loca-tion at study entry or at time of diagnosis is often used toestimate lifetime or long-term exposure estimates in epi-demiological studies. Given that approximately half of allindividuals move within a five year period [7] and thatresidential mobility varies depending on socio-economicfactors [8-11], there is potential for exposure misclassifica-tion and bias in studies that ignore or improperly accountfor residential mobility. While there is growing recognitionof the need for spatiotemporal epidemiology approachesand life-time residential histories in exposure assessment[12], mainly in cancer epidemiology [13,14], little is knownregarding the potential exposure misclassification and biasresulting from self-reported residential histories, the mostcommon form of attaining residential histories in epide-miological studies [15], and from the assumption of resi-dential stationarity in air pollution epidemiology.Incorporating residential histories into air pollutionexposure assessments requires corresponding air pollutionconcentration estimates that cover the spatiotemporaldomain of the study period. To date, the associationbetween air pollution and lung cancer has been examinedusing a variety of study periods and exposure assessmentapproaches. The most common approaches have aggre-gated air pollution monitoring levels within cities ordefined areas [1,2,6,16], estimated ambient air pollutionlevels at residential addresses using fixed-site monitoringdata or dispersion models [3-5,17,18], or used proximityto roads and industrial sources as exposure surrogates[19,20]. In terms of national retrospective exposure assess-ment studies, few are available that examine multiple pol-lutants and exposure sources [21,22].Here we develop a comprehensive spatiotemporalexposure assessment approach for Canada and apply it toa population-based case-control study of 8353 individualswho provided lifetime self-reported residential histories.For the exposure period 1975 to 1994, we assign fine par-ticulate matter (PM2.5), nitrogen dioxide (NO2) andozone (O3) air pollution exposures, as well as exposuresto vehicle and industrial emissions. The implications ofdisregarding and/or improperly accounting for residentialhistories in long-term exposure assessments are alsoexamined. The exposure assessment methods developedproduce annual spatiotemporal exposure estimates andwill allow subsequent epidemiologic analyses to examinelatency periods, to include both urban and rural popula-tions, and to study the contributions of multiple ambientpollutants and local vehicle and industrial emissions tolung cancer risk in Canada.MethodsThe lung cancer case-control studyWe utilize the lung cancer component of the NationalEnhanced Cancer Surveillance System (NECSS), whichincludes 3280 histological-confirmed lung cancer casesand 5073 population controls collected between 1994 and1997 in the provinces of British Columbia, Alberta, Sas-katchewan, Manitoba, Ontario, Prince Edward Island,Nova Scotia and Newfoundland. The respective ethicsreview boards of each province reviewed and approved theNECSS study. Due to residential mobility, study partici-pants are located in all provinces of Canada requiringnational-level exposure assessment. Johnson et al. [23]describe the overall recruitment methodology for theNECSS. Briefly, cases were identified through provincialcancer registries and mailed a research questionnaire. Theresponse rate for contacted lung cancer cases was 61.7%.Population controls were selected from a random sampleof individuals within each province, with an age/sex distri-bution similar to that of all cancer cases (strategies forrecruiting population controls varied by province depend-ing on data availability and accessibility). Provincial cancerregistries collected information from sampled controlsusing the same protocol as for the cases. The responserate for contacted population controls was 67.4%.Residential histories at the 6-digit postal code level arethe basis of the air pollution exposure assessment reportedhere. In urban areas a 6-digit postal code typically incorpo-rates one side of a city block, but represent substantiallylarger areas in rural locations (e.g. greater than 100 km2 inremote locations of Canada). Residential histories wereconverted to postal codes by the Public Health Agency ofCanada and geocoded using DMTI Inc. 1996 postal codes.While lifetime residential histories were collected, theexposure period was restricted to 1975 to the start ofstudy enrolment (1994), due to the presence of recall biasin earlier reported histories (explained in more detail inthe discussion section) as well as the lack of informationon postal code locations, air pollution monitoring dataand geographic information prior to 1975.Air pollution exposure assessment approachA multi-staged approach was required to assign ambientair pollution concentrations to residential histories from1975 to 1994. The spatiotemporal exposure assessmentincluded three steps. First, national spatial surfaces werecreated from recent satellite-based estimates (for PM2.5Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 2 of 13and NO2) and a chemical transport model (for O3). Sec-ond, all National Air Pollution Surveillance (NAPS) mon-itoring data were compiled and formatted for the studyperiod, including 120 NO2 stations and 1030 measure-ment-years, 187 O3 stations and 1440 measurement-years, 177 TSP stations and 1826 measurement-years,and 25 PM2.5 stations and 141 measurement-years. Dueto the small number of PM2.5 measurements available,and no measurements made prior to 1984, a randomeffect model was used to estimate PM2.5 based on TSPmeasurements and metropolitan indicator variables.Finally, the spatial pollutant surfaces were calibratedyearly to estimate average annual concentrations between1975 and 1994. Two approaches were used for calibra-tion: the first estimated historical annual averages usingsmoothed inverse distance weighting (IDW) interpolationof the ratios of spatial co-located historical NAPS andsurface estimates, while the second used linear regressionmodels.Exposure to vehicle emissions was estimated usingproximity to highways and major roads, adjusted basedon historical vehicle emissions in Canada. Exposures toindustrial emissions were calculated based on proximityto major and minor industrial sources extracted from acomprehensive database of industrial facilities in Canadaoperating during the study exposure period. Estimatesfor different vehicle and industrial emission sources werenot converted into concentrations and added to ambientconcentration estimates as we want to examine eachsource and distance threshold separately in subsequentepidemiological analyses. Specific components of theexposure assessment approach are described in detailbelow.National spatial pollutant surfacesSpatial models of ambient PM2.5, NO2 and O3 concentra-tions were developed to represent current spatial pollutionpatterns across Canada. A PM2.5 surface was derived fromAerosol Optical Depth (AOD), using data from the Mod-erate Resolution Imaging Spectroradiometer (MODIS) andthe Multiangle Imaging Sectroradiometer (MISR) satelliteinstruments, and was combined with a chemical transportmodel (GEOS-Chem; http://www.geos-chem.org) to esti-mate the relationship between aerosol optical depth andsurface PM2.5 (for full details see [24]). Estimates for PM2.5represented a composite estimate developed from 2001 to2006 and included locations with greater than 100 validmeasurements to ensure estimate representativeness. TheNO2 surface was estimated from tropospheric NO2 col-umns retrieved from the Ozone Monitoring Instrument(OMI) and also used GEOS-Chem to calculate the rela-tionship between the NO2 column and surface NO2 [25].NO2 estimates used data from 2005 to 2007 as OMI mea-surements began in late 2004. Both PM2.5 and NO2 wereestimated at a 0.1 × 0.1 degree resolution (~10 × 10 km).The O3 surface was created from the Canadian Regionaland Hemispheric O3 and NOx System (CHRONOS) [26].This model is reinitialized every 24 h with meteorologyand is fused with the O3 observations across Canada andthe U.S. on an hourly basis using an optimal interpolationapproach based upon a least square combination of theCHRONOS and measured O3 data that minimized theerror variance. This surface was created at a 21 km resolu-tion and represents average summer (May through Sep-tember) concentrations from 2004 to 2006. Figure 1illustrates the PM2.5, NO2 and O3 pollutant surfaces usedto represent current spatial concentrations across Canada.Next, these surfaces were calibrated with NAPS monitor-ing data to estimate historical annual spatial exposuresurfaces.Air pollution monitoring dataThe NAPS monitoring network began measurements ofTSP in 1970, NO2 and O3 in 1975 and PM2.5 and PM10in 1984. Figure 2 illustrates the location of all NAPSmonitors in Canada, 1975 TSP monitoring stations with50 km buffers (for reference of historical monitor spatialcoverage) and all study participant residential postalcodes between 1975 and 1994.NAPS monitoring data were first formatted intomonthly averages for all pollutants. Continuous monitor-ing data were included if at least 50% of daily hourlyobservations were available and at least 50% of days wereavailable in a month. Monthly averages from dichotomoussamplers (PM2.5) required a minimum of 3 of 5 validmonthly measurements. Yearly averages were not calcu-lated unless there were at least six months of completedata with one month per season, and summer O3 averagesunless there were 3 months of data available. Supplemen-tal material, Figure1 illustrates historical annual averagepollutant concentrations from available NAPS monitoringstations that were in operation for all years. Temporaltrends show a large decrease in TSP concentration duringthe study period (51% from 1970 to 1994), a decrease inNO2 (28% from 1975 to 1994) and PM2.5 (32% from 1984to 1994), and an increase in O3 (19% from 1975 to 1994).Importantly, the changes in pollutant concentrations werenot uniform across geographic areas in Canada.Modeling historical PM2.5 concentrations from TSPDue to the lack of historical spatial and temporal PM2.5measurement coverage, we used co-located PM2.5 andTSP measurements between 1984 and 2000 to create pre-dictive models of historical PM2.5 concentrations. Theoverall approach to estimating PM2.5 is similar to thatused by Lall et al. [27] to estimate metropolitan area spe-cific PM2.5 and PM10 relationships with TSP across theU.S. We used random effect models (GLIMMIX proce-dure in SAS 9.3) to account for the clustering of annualHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 3 of 13measurements over time at each NAPS station. Table 1summarizes the final PM2.5 model incorporating TSPconcentrations (μg/m3) and census metropolitan area(CMA) indicator variables. The R2 and RMSE for thePM2.5 model was 0.67 and 2.31. Figure 3 illustrates themeasured and predicted PM2.5 concentrations. Theresulting PM2.5 model was applied to all valid TSP moni-toring stations; the nearest CMA core within 100 km wasused to determine the CMA model coefficient for thePM2.5 model, otherwise no CMA variable was includedin the model. Figure 2 in the supplemental material mapsthe CMA’s used in the model and areas covered by the100 km buffers.Calibrating spatial pollutant surfaces using historical dataTwo approaches were used to extrapolate current PM2.5,NO2 and O3 surfaces to estimate annual concentrationsbetween 1975 and 1994. Both approaches were devel-oped using 90% of the monitoring data available foreach year, while retaining 10% for model evaluation.Model performance was assessed using adjusted R2 androot-mean-square error (RMSE).The first approach calibrates the current spatial sur-faces (shown previously in Figure 1) using annual NAPSmonitoring data and smoothed IDW interpolation ofthe ratio’s of spatial co-located historical NAPS and sur-face estimates. The yearly calibrations were performedusing the following equation:Yearly Historical Surfacej = Surfacex,y×⎡⎢⎢⎢⎢⎢⎢⎣Nnapsk=1⎛⎝ 1(dx,y,k)x NAPSJkSurfacek⎞⎠napsk=11(dx,y,k)⎤⎥⎥⎥⎥⎥⎥⎦(1)Where for each year between 1975 and 1994 the annualhistorical surface for pollutant j is equal to the currentspatial surface of pollutant j (Surfacex,y) at coordinates x,ymultiplied by the IDW interpolation of the ratio’s of spa-tial co-located historical NAPS and surface estimates. dx,y,k is the distance (km) from NAPS monitoring station kto location x,y. NAPSJK and Surfacek are coincidentlysampled pollutant concentrations of j at station k. Asmooth interpolation option (smooth factor = 0.2) wasincluded in the IDW interpolation (not shown in equa-tion 1 for simplicity), which uses three ellipses in theinterpolation method: points that fall outside the smallerellipse but inside the largest ellipse are weighted using asigmoid function [28]. The smoothed IDW function wasused to reduce abrupt changes in the yearly calibrationsurfaces as these do not reflect spatial patterns of pollu-tion change.The second approach uses linear regression to modelannual concentrations. Predictor variables include thespatial pollutant surfaces, a time-trend and historicalpopulation density data. Population location data werederived from the 1971, 1976, 1981, 1986, 1991, andFigure 1 National pollutant surfaces created from recentsatellite estimates (for PM2.5 and NO2) and a dispersion model(for O3). Insets represent higher population density locations inCanada (south western BC and southern Ontario and Quebec).Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 4 of 131996 Canadian census; between census years wereassigned the nearest census. The annual population den-sity variables were calculated in a GIS for various bufferdistances (1 km to 50 km’s) around each NAPS monitor.Roads and industry were not included in the models aswe want to separately evaluate exposure to these sourcesand lung cancer risk. We used random effect models(GLIMMIX procedure in SAS 9.3) to account for theclustering of annual measurements over time at eachNAPS station and selected predictor variables that maxi-mized model fit. We estimated R2 and RMSE statisticsby predicting the measurement data with the fixed-effectcoefficients using ordinary least squares regression.Exposure to vehicle emissionsExposures to vehicle emissions were estimated usingproximity measures to highways (freeways and majorhighways) and major roads (freeways, highways, andarterial and collector roads). The 1996 DMTI Inc. roadnetwork was used to derive proximity measures for allcase and control residential years, due to the lack of his-torical national road networks. The average distance toeach road class was calculated separately as well as thenumber of years residing within 50, 100 and 300 m of ahighway and/or major road. These proximity distanceswere selected as vehicle related pollutant gradients, suchas for NO2 and volatile organic compounds, are highestwithin 50 and 100 m of a major road but remain signifi-cantly elevated to 300 m [29].Emissions from vehicles have changed significantlyover time due to increases in vehicle kilometres travelledand improved vehicle emission controls [30,31]. Exposureindicators for years residing near highways and majorroads were therefore weighted to account for theseFigure 2 Location of all national air pollution surveillance monitors in Canada and study participant residential postal codes between1975 and 1994.Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 5 of 13changes. Supplemental material, Figure 3 shows thedecrease in the total NOx emissions from on-road mobilesources in Canada (used here to represent primaryvehicle emissions), including heavy and light duty dieseland gasoline vehicles, from 1980 to 2007 and extrapo-lated levels to 1970. NOx emissions estimates were com-piled by Environment Canada using the latest emissionestimation methodologies and statistics available as ofMarch 2008. Emission factors were developed usingMOBILE6.2 C and the number of vehicle kilometres tra-velled. MOBILE6.2 C is a vehicle emissions modelingsoftware specific to Canada and accounts for the vehiclefleet profile, vehicle emission standards, and fuel charac-teristics [32]. Given the NOx emissions trends documen-ted in the United States from 1970 to 1980 [33], linearextrapolation was used to estimate NOx emissions from1980 to 1970. The ratio of resulting 1994 and 1975 NOxemission estimates suggest that living near a major roadin 1975 is equivalent to 1.26 “1994” years due to changesin vehicle emissions (the ratio also accounts for changesin vehicle numbers). A weighting factor (1 + 0.013*(1994-proximity exposure year)) was therefore used to adjustproximity-based vehicle exposures to account fordecreases in the magnitude of vehicle emissions over thestudy period.Exposure to industrial emissionsA comprehensive inventory of industrial emissionssources was compiled as part of the NECSS within theEnvironmental Quality Database (EQDB) [23,34,35].Locations of industrial manufacturing facilities and activ-ities in approximately fifty standard industrial classifica-tions (SIC) from 1970 to 1994 are included in thedatabase along with operational time periods. Approxi-mately 7800 sources with a 4 digit SIC are included and8200 municipal waste facilities. Major industries, includ-ing metal smelters, pulp and paper mills, petroleum pro-duct companies, foundry and steel plants, aluminumsmelters, non-hydro power plants, and petrochemicalcompanies, contain pollutant discharge estimates whileminor industrial sources have no emission records. Thedistance between an industrial source and a subjects’postal code has been validated to +/-150 m in urban loca-tions [34]. The EQDB has been used in conjunction withthe NECSS to examine leukemia and chlorination by-products [36] and residential proximity to industrialplants and Non-Hodgkin’s Lymphoma [37]. We calculateexposure to major industrial emissions and to minorsources within 1, 2 and 3 km buffers from residentialpostal codes. These distances were selected to ensurespecificity of proximity based exposure assessments formultiple industries and substances. Similar distancethresholds have been used previously in small area healthstudies [38,39]. To be considered exposed, and to calcu-late the number of years exposed to each proximity cate-gory, at least 1 industrial facility had to be operatingwithin the associated buffer distance.Table 1 Model used to predict historical PM2.5 using TSPmeasurements and census metropolitan area indicatorvariables (R2 = 0.67, RMSE = 2.31).Variables Estimate SE pIntercept 1.93 2.30 0.42TSP 0.13 1.78e-2 < 0.001*CMA IndicatorCalgary 0.44 2.63 0.87Edmonton -1.82 2.69 0.50Halifax 7.71 3.02 0.01*Hamilton 4.76 3.02 0.12Montreal 6.01 2.42 0.01*Ottawa 4.86 2.94 0.10Quebec 3.17 2.60 0.23St. Johns 5.72 3.81 0.13Saint John 3.28 30.7 0.29Toronto 5.63 2.60 0.03*Vancouver 6.50 2.47 0.01*Victoria 2.48 2.73 0.36Windsor 5.63 2.56 0.03*Winnipeg 1.00 - -Model performance: R2 = 0.67, RMSE = 2.31.R2 and RMSE estimated byregressing the predictions from the fixed-effects terms against measuredvalues*Significant at p < 0.05Figure 3 Correspondence between predicted PM2.5concentrations using TSP concentrations and metropolitanindicator variables and NAPS PM2.5 measurements.Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 6 of 13ResultsResidential historiesThe NECSS questionnaire asked participants to list eachplace in Canada that they had lived for at least one year.A total of 8176 individuals (98%) reported at least onefull 6-digit postal code and 6918 individuals (83%)reported at least 15 years of residential histories from1975 to 1994. On average, individuals reported 2.3 (SD =1.6) different residences from 1975 to 1994; 1617 indivi-duals lived only in rural areas and 4222 individuals livedonly in urban areas of Canada. Urban areas were definedusing Statistics Canada community size classifications(urban core, urban fringe, urban areas outside of CMA,rural fringe, and rural areas outside of CMA). In total,77% of the studies exposure-years occurred in urbanareas.Importantly, while no significant difference (p = 0.54)was found in the number of geocoded residential-yearsbetween cases and controls for the 1975 to 1994 exposureperiod, cases tended to report older addresses more oftenthan controls. Recall bias was especially evident for resi-dential histories prior to 1975, as shown in Figure 4.Ambient exposure assessmentsThe first approach to calibrating current pollution surfacesused IDW interpolation to create annual surfaces between1975 and 1994. Figure 5 illustrates the resulting PM2.5exposure surfaces for 1975, 1980, 1985, 1990 and 1994,PM2.5 measurement locations with 50 km buffers, theaverage PM2.5 exposure surface between 1975 and 1994,and the location of the case-control study subjects. Twentyannual exposure surfaces were created from 1975 to 1994,but only five are shown here. The study population resi-dential years indicates the locations of all yearly residentialhistories during the twenty year exposure period summedwithin a 50 km grid. The temporally adjusted surfaces forNO2 and O3 are provided in Figures 4 and 5 of the supple-mental material.The performance of the linear regression models wasmoderate for all three pollutants (PM2.5 R2 = 0.33, NO2R2 = 0.36 and O3 R2 = 0.47) as described in Table 2.Population density within 10 km of monitoring stationswas most strongly associated with PM2.5, while popula-tion density with 5 km was most strongly associated withNO2 (positively associated) and O3 (negatively asso-ciated). A linear time-trend did not improve the O3model and was therefore not included in the final model.Evaluation of the two historical calibration approachesare shown in Table 3 which summarizes the R2 andRMSE of model evaluations using the 10% sample ofmonitoring data withheld each year. The spatiotemporalIDW interpolation of PM2.5 had the best performance(R2 = 0.51), while the NO2 and O3 linear models had thebest performance (R2 = 0.38 and R2 = 0.56). Model per-formance tended to decrease for older measurements,but not substantially. Additional file 1: Supplementalmaterial 1, Figure 6 presents the scatter plots for eachmodel evaluation.Table 4 presents the exposure assessment results usingboth historical calibration methods and air pollutionexposures derived from NAPS monitoring data within 50km of residential postal codes. To ensure accurate expo-sure assessment, results are presented for individualswith at least 15 complete exposure-years between 1975and 1994. Exposures for different time-periods (e.g.1975-1980, 1975-1985, and 1975-1990) were also calcu-lated to examine different latency periods (data notshown).Exposure to vehicle and industrial emissionsProximity measures used to represent exposure to vehicleemissions are summarized in Table 5. Individuals livedwithin 50, 100 and 300 m of a highway for a mean of 0.5(SD = 2.9), 1.1 (SD = 4.0) and 2.9 (SD = 6.3) years,respectively. Exposure years increased slightly whenweighted by temporal emission changes. The averagemean distance from study participants’ postal codes tothe nearest highway was 3.9 km. When residential his-tories were restricted to urban areas (where proximity isa more accurate measure of exposure than in ruralareas), the distance to highways and major roadsdecreased substantially. Over half of the study populationwas exposed to emissions from a major road at somepoint during the study period (i.e. had lived at least oneyear within 300 m of a major road).The number of years study participants lived within 1,2 and 3 km of a major and minor industry are summar-ized in Table 6 as are aggregated emission estimates forFigure 4 Percent of cases and controls reporting residentialaddresses at the 6-digit postal code level from the start ofstudy enrollment (1994) to1944.Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 7 of 13major industrial sources. Proximity to specific emissionsources (e.g. oil refineries, smelters, and pulp and papermills) were also calculated (data not shown). Individualslived within 1, 2 and 3 km of a major industrial sourcefor a mean of 1.6 (SD = 5.3), 4.3 (8.3) and 6.4 (9.5)years respectively. Over half of the study population (n= 5942) lived within 3 km of a minor industrial sourcefor at least one year between 1975 and 1994.Disregarding residential histories and exposure errorA total of 3305 study participants (40%) lived at theirstudy entry address for the entire twenty year exposureperiod, while 622 (7.6%) participants lived for 15-19years, 970 (11.9%) for 10-14 years, 1433 (17.5%) for 5-9years, and 1756 (23%) for less than 5 years. Correlationbetween ambient air pollution exposures derived fromstudy entry residential addresses only, in place ofexposures derived from residential histories and spatio-temporal air pollution models, were relatively high forPM2.5 r = 0.70, NO2 r = 0.76 and O3 r = 0.72. However,when examining exposure misclassification based onincorrectly assigned exposure quintiles, 50%, 49% and46% of individuals where classified into a differentPM2.5, NO2 and O3 quintile. When temporal variation isremoved from the exposure assessment (i.e. historicalexposures are derived from residential histories appliedto the current spatial pollution surfaces) 17%, 15% and14% of individuals where classified into a differentPM2.5, NO2 and O3 exposure quintile. Similar resultswere found for proximity based exposures, for example,30% of individuals classified as not exposed to highwayemissions based on their address at study entry wereactually exposed when residential histories were usedfor exposure assessment.Figure 5 Example of annual PM2.5 exposure surfaces created using the IDW interpolation calibration approach for all years between1975 and 1994.Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 8 of 13DiscussionIncorporating residential mobility in chronic air pollutionstudies is fundamental to accurate exposure estimates.Boscoe [15] presents a review of environmental healthstudies that have incorporated residential histories to-date. In our study, only 40% of participants lived at theirstudy entry residence for the entire 20 year exposure per-iod; on average, 2.3 (SD = 1.6) different residences persubject were reported. Recall bias was present for self-reported residential histories prior to 1975, with casesrecalling older residences more often than controls. Thishas important implications for environmental epidemiol-ogy using self-reported residential histories as manyenvironmental exposures have decreased substantiallyover time. Consequently, exposure assessment based on agreater proportion of older residential histories in casescompared to controls will result in an upward bias, ratherthan non-differential bias typically assumed from expo-sure misclassification. Studies that incorporate self-reported residential histories, particularity long-term resi-dential histories - in this case over twenty years, mayneed to account for reporting bias in epidemiologicalanalysis.This study also demonstrated the importance of esti-mating air pollution exposures from residential histories,both in terms of including different residential locationsas well as the corresponding spatiotemporal air pollutionconcentration estimates. Exposure quintiles based onresidential addresses at study entry had approximately50% correspondence to exposure quintiles developedfrom residential histories and spatiotemporal air pollu-tion surface. These results address one of the researchopportunities suggested by Meliker and Sloan [12]:“indentifying circumstances under which it is worth-while to compile and incorporate extensive space-timedata histories of mobility or environmental contami-nants”. Epidemiological studies of diseases with longlatency periods (in this case lung cancer) and/or thatexamine spatially and temporally varying exposures (inthis case ambient air pollution) are clearly suchcircumstances.Table 3 Evaluation of spatiotemporal IDW interpolationand linear regression models to predict annual historicalair pollution.IDWInterpolationLinearModelsYear Stations N R2 RMSE R2 RMSENO2 All 120 1030 0.22 6.66 0.38 5.921994-1990 94 349 0.30 5.66 0.36 5.421989-1985 88 300 0.20 6.61 0.44 5.541984-1980 62 226 0.13 6.72 0.40 5.621979-1975 52 155 0.17 8.75 0.29 8.07PM2.5 All 177 1826 0.51 2.96 0.30 3.531994-1990 106 446 0.64 1.96 0.32 2.701989-1985 113 480 0.57 2.30 0.36 2.811984-1980 124 476 0.34 3.79 0.12 4.361979-1975 123 424 0.43 3.32 0.26 3.77O3 All 187 1440 0.39 5.29 0.56 4.481994-1990 158 582 0.53 4.92 0.65 4.251989-1985 125 409 0.36 5.41 0.54 4.571984-1980 80 286 0.25 4.67 0.28 4.571979-1975 48 163 0.22 6.33 0.60 4.50Table 4 Ambient exposure estimates derived from NAPSmonitors within 50 km of residential postal codes andspatiotemporal exposure models.Pollutant N* Mean SD Min IQR MaxNAPS Measurements ≤ 50 kmTSP (μg/m3) 4027 60.0 16.9 22.3 21.4 114.1Modeled PM2.5 (μg/m3)a 4027 17.0 2.5 11.9 3.4 25.7NO2 (ppb) 3649 23.4 6.0 6.0 7.6 37.8O3 (ppb)b 4382 21.0 3.9 7.0 5.3 32.6Spatiotemporal IDW InterpolationPM2.5 (μg/m3) 6833 11.3 2.6 3.6 3.9 19.0NO2 (ppb) 6919 15.3 8.8 1.1 14.5 43.4O3b(ppb) 6919 23.2 3.7 12.9 4.6 35.4Linear Regression ModelsPM2.5 (μg/m3) 6833 9.1 1.9 4.7 2.2 16.1NO2 (ppb) 6919 17.7 4.1 13.1 5.0 35.1O3b (ppb) 6919 26.4 3.4 18.1 4.7 37.2*Number of individuals with ≥ 15 complete exposure-yearsa Modeled using TSP and CMA indicator variables as described previously inTable 1b Summer (May through September) O3Table 2 Results of historical PM2.5, NO2 and O3 linearregression models.Model Distance Value SE pPM2.5 Model [R2 = 0.33, RMSE = 3.57]Intercept - 1.18 1.16 0.31Satellite PM2.5 - 0.46 0.11 < 0.001Population Density 10 km 3.94e-6 2.89e-7 < 0.001Years < 1994 - 0.29 9.28e-3 < 0.001NO2 Model [R2 = 0.36, RMSE = 7.00]Intercept - 10.88 1.07 < 0.001Satellite NO2 - 1.67 0.46 < 0.001Population Density 5 km 2.6e-5 5.11e-6 < 0.001Years < 1994 - 0.28 0.028 < 0.001O3 Model [R2 = 0.47, RMSE = 5.13]Intercept - 6.85 1.66 < 0.001O3 Dispersion Model - 0.73 0.06 < 0.001Population Density 5 km -2.0e-5 2.5e-6 < 0.001Hystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 9 of 13Despite the fact that the Canadian NAPS monitoringnetwork is one of the longest-standing national air pollu-tion monitoring programs worldwide and now covers themajority of urban centers in Canada, its limited spatiotem-poral coverage necessitated the creation of national mod-els that capture both urban and rural populations. Wewere able to use NAPS data within 50 km of residentialpostal codes to assign exposures to 63%, 70% and 54% ofexposure-years for TSP, O3 and NO2. Very limited spatialand temporal PM2.5 monitoring data were available (only40% of exposure-years between 1984 and 1994 could beassigned) and we therefore estimated historical PM2.5using TSP and metropolitan area indicator variables. Theresulting models predicted PM2.5 variability well; the ratiofor modelled PM2.5/TSP (0.32, SD = 0.12) is very similarto that found in US metropolitan areas (PM2.5/TSP = 0.30,SD = 0.11) [27].National spatial pollutant surfaces were compiled andcalibrated with historical NAPS data to assign ambientpollutant concentrations to all study participants’ residen-tial postal codes between 1975 and 1994. The twoapproaches used to calibrate spatial pollutant surfaces dif-fer in their approach to account for temporal and spatialchange; IDW interpolation accounted for the heterogene-ity in pollution level changes across Canada duringthe exposure period, while linear regression modelsincorporated a linear time-trend and population density asa spatial predictor. The interpolation approach betterrepresented historical PM2.5 concentrations, potentiallydue to the larger spatial scale of PM2.5, while the linearregression models better represented historical NO2 andO3 concentration, which have finer spatial resolutions.The creation of national spatiotemporal modelsallowed for the inclusion of all study participants, regard-less of geographic location and NAPS monitor coverage.This was important as 42884 (23%) of exposure-yearsoccurred in rural areas. The mean PM2.5, NO2 and O3exposure estimates derived from the spatiotemporalmodels were 11.3 μg/m3 (SD = 2.6), 17.7 ppb (4.1), and26.4 ppb (3.4) respectively. The magnitude of these expo-sures are less than those used in other studies, for exam-ple, the widely cited ACS study (PM2.5: 17.7 μg/m3 (3.0),NO2 21.4 ppb (7.1); and O3 45.5 ppb (7.3)) [1]. This islikely due to the inclusion of rural study participants aswell as lower ambient pollution levels in Canada. Theability to incorporate rural areas in the exposure assess-ment added to the variability in the studies exposure esti-mates, particularly for NO2 and O3, as the majority ofhistorical NAPS measurements in Canada represent pol-lutant concentration in large urban areas.The results of the retrospective air pollution modelingapproach conducted here are comparable to other suchTable 5 Proximity measures to highways and major roads.Proximity Measure # of People Exposeda # of Years Exposed (Mean ± SD) # of Weightedb Years Exposed (Mean ± SD)Highways≤ 50 m 341 0.5 (2.9) 0.7 (3.9)≤ 100 m 647 1.1 (4.0) 1.5 (5.4)≤ 300 m 1640 2.9 (6.3) 4.0 (8.5)Major Roads≤ 50 m 1438 2.3 (5.5) 3.2 (7.6)≤ 100 m 2283 4.0 (6.9) 5.5 (9.5)≤ 300 m 4517 10.1 (8.8) 13.8 (12.1)a Number of individuals living > 1 year within 50/100/300 m of a highway or major roadb Weighted to account for temporal changes in vehicle emissionsTable 6 Proximity measures to major and minor industrial sources.Proximity Measure # of People Exposeda # of Years Exposed (Mean ± SD) # of Facilities (Mean ± SD) Emissionsb (tonnes) (Mean ± SD)Major Industries≤ 1 km 838 1.6 (5.3) 6.2 (5.5) 4.5e5 (3.6e7)≤ 2 km 1995 4.3 (8.2) 13.3 (11.6) 4.5e5 (3.5e7)≤ 3 km 2743 6.4 (9.5) 21.3 (18.6) 1.9e3 (1.6e4)Minor Industries≤ 1 km 4137 11.4 (11.2) 32.6 (59.3) -≤ 2 km 5515 16.7 (10.0) 115.7 (163.2) -≤ 3 km 5942 18.9 (9.0) 218.0 (303.8) -a Number of individuals living > 1 year within 1/2/3 km of a major or minor industrial sourceb Summary of facility emissions > 0 tonnes. Only available for major industriesHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 10 of 13studies; however, the majority of retrospective air pollu-tion exposure assessments have been conducted solelyfor urban areas. For example, Bellander et al. [18] usedemission data, dispersion models, and geographic infor-mation systems (GIS) to assess exposure to NO2, NOxand SO2 ambient air pollution during 1960, 1970 and1980 in Stockholm, Sweden. Model evaluation using his-torical data was not possible, but the model was foundto have high correlation (r = 0.96) with aggregated1994-1997 data from 16 monitors. In terms of nationalmodels, Hart et al. [22] developed U.S. nationwide mod-els of annual exposure to PM10 and NO2 from 1985 to2000. Generalized additive models were used to predictspatial surfaces from monitoring data and GIS-derivedcovariates (e.g. distance to road, elevation, proportion oflow-intensity residential, high-intensity residential, andindustrial, commercial land use). Model performance(R2) for PM10 and NO2 was 0.49 and 0.88 respectively.Another national retrospective study was conducted aspart of the Netherlands Cohort Study on Diet and Can-cer [21]. Ambient air pollution exposures were esti-mated using regional (IDW monitor interpolation),urban (regression modelling), and local (road proximity)components. This approach explained 84%, 44%, 59%and 56% of the variability in averaged monitor databetween 1976 and 1997 for NO2, NO, BS and SO2,respectively. The density of monitors in the Netherlandsand the use of aggregated monitoring data may explainthe higher model performance than seen in this study.The exposure assessment approach presented here capi-talizes on study participants’ lifetime residential historiesand incorporates comprehensive modelling approaches toestimate exposures to ambient air pollution and to vehicleand industrial emissions. Nevertheless, there are severallimitations to this approach that may lead to exposuremisclassification. Due to privacy concerns, residentialaddresses were coded using a standard geographic refer-ence of 6-digit postal codes. Using a set geographic refer-ence reduced error from changing postal codes over time;however, the spatial accuracy of postal codes varies sub-stantially between urban and rural areas of Canada. Proxi-mity analyses for exposures to vehicle and industrialemissions will therefore be more accurate in urban areas.The ambient air pollution exposure assessment relies onthe accuracy of NAPS monitoring data, and historicalmonitor locations, especially in rural areas, may have beensited to capture local pollution problems. Unfortunately,no historical data were available to evaluate the represen-tativeness of NAPS monitoring data. Due to sparse tem-poral and spatial PM2.5 monitor coverage, we createdhistorical models based on TSP monitoring data andCMA indicator variables. While the model had good pre-diction, it was created from a limited number of monitor-ing stations from 1984 to 2000. Nevertheless, severalstudies have estimated PM2.5 successfully from TSP [6,27].The accuracy of the final spatiotemporal PM2.5, NO2 andO3 surfaces is also determined from the initial concentra-tion surface as well as fusion with historical NAPSmonitoring data or predictions incorporating a lineartime-trend and population density. Some anomalies existin the current spatial surfaces, for example, high PM2.5concentrations in mountainous regions and PM2.5 andNO2 in certain locations in the Prairies; however, fewstudy participants lived in these locations and exposuremisclassification is therefore limited. All historical moni-tors were used to adjust annual spatial pollution surfaces,which resulted in urban monitor ratios extrapolated torural areas. Few rural monitors exist and it was not possi-ble to restrict to rural monitors when adjusting the spatialpollution surfaces in rural areas. Exposure to vehicle emis-sions was based on proximity measures to a national 1996road network and a clear limitation was the lack of histori-cal road databases. Industrial emissions were based on acomprehensive database of industrial locations from 1970to 1994; however, emission estimates were only availablefor major industries, which restricted the examination ofspecific industrial chemicals when minor industries wereincluded.ConclusionsWe conducted a comprehensive air pollution exposureassessment for a population based lung cancer case-controlstudy of 8353 individuals using self-reported residentialhistories between 1975 and 1994. Incorporating residentialhistories was an important component of the exposureassessment approach, and necessitated the creation ofnational spatiotemporal air pollution models. Due to thelack of historical air pollution measurements, as well as dif-ferences in data availability between urban and rural areas,a number of modelling approaches were used to assignannual ambient PM2.5, NO2 and O3 concentrations, as wellas proximity measures for vehicle and industrial emissions,to study participants’ residential addresses. The exposureassessment methods developed here will allow subsequentepidemiological analyses to examine latency periods asso-ciated with lung cancer, include both urban and ruralpopulations, and study the contributions of multiple ambi-ent pollutants and local vehicle and industrial emissions tolung cancer risk in Canada. In addition, this exposureassessment has demonstrated the importance of includingresidential histories in long-term exposure assessments, aswell as the need to carefully examine self-reported residen-tial histories for recall bias.Additional materialAdditional file 1: Supplemental material: Figure 1 Annual average (SD)pollutant concentrations from all valid historical NAPS monitoringHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 11 of 13stations that were operating for the entire study period. Figure 2 CensusMetropolitan Areas (CMA’s) in Canada with PM2.5 and TSPmeasurements used to create predictive models of historical PM2.5concentrations. Figure 3 Yearly NOx on-road mobile emissions in Canadafrom 1980 to 2007 and extrapolated levels to 1970. Figure 4 NO2exposure surfaces (note: 20 annual surfaces were created but only 5 areshown here) and locations of NAPS monitors with 50 km buffers. Thestudy population residential years represents all residential locationsbetween 1970 and 1994 summed within a 50 km grid. Figure 5 O3exposure surfaces (note: 20 annual surfaces were created but only 5 areshown here) and locations of NAPS monitors with 50 km buffers. Thestudy population residential years represents all residential locationsbetween 1970 and 1994 summed within a 50 km grid. Figure 6 Scatterplots of measured versus predicted PM2.5, NO2 and O3 for IDWinterpolation and linear regression models.AbbreviationsPM2.5: Fine Particulate Matter; NO2: Nitrogen Dioxide; O3: Ozone; TSP: TotalSuspended Particulates; PM10: Course Particulate Matter; NECSS: NationalEnhanced Cancer Surveillance System; NAPS: National Air PollutionSurveillance; IDW: Inverse distance weighting; AOD: Aerosol Optical Depth;MODIS: Moderate Resolution Imaging Spectroradiometer; MISR: MultiangleImaging Sectroradiometer; GEOS-Chem: Chemical transport model; OMI:Ozone Monitoring Instrument; CHRONOS:: Canadian Regional andHemispheric O3 and NOx System; U.S.: United States; CMA: CensusMetropolitan Area; EQDB: Environmental Quality Database; SIC: StandardIndustrial ClassificationsAcknowledgementsWe would like to thank: the Canadian Cancer Registries EpidemiologicResearch Group for the lung cancer case-control data; the National AirPollution Surveillance (NAPS) program for the air pollution monitoring data;and Qian Li and Ilan Levy for helping create the O3 spatial surface. PH issupported a UBC Bridge scholarship, a Michael Smith Foundation for HealthResearch senior graduate trainee award, and a Canadian Institute of HealthResearch Frederick Banting and Best research scholarship.Author details1School of Population and Public Health, University of British Columbia, 2206East Mall, Vancouver, BC V6T 1Z3, Canada. 2Occupational Cancer ResearchCentre, Cancer Care Ontario, Ontario, Canada. 3Science Integration Division,Centre for Chronic Disease Prevention and Control, Public Health Agency ofCanada, Ontario, Canada. 4Air Quality Research Division, Environment,Ontario, Canada. 5Department of Physics and Atmospheric Science,Dalhousie University, Ontario, Canada. 6Atmospheric Chemistry andDynamics Branch, NASA Goddard Space Flight Center, Greenbelt, USA.7Department of Physics and Atmospheric Science, Dalhousie University,Canada; Harvard-Smithsonian Center for Astrophysics, Cambridge, USA.8School of Population and Public Health, University of British Columbia,Vancouver, BC, Canada.Authors’ contributionsPH, PAD and MB designed and implemented the air pollution exposureassessment approach; KCJ implemented the NECSS case-control study; JBcreated the O3 spatial surface; and AVD, LL and RM created the PM2.5 andNO2 spatial surface. All authors have read and approved the finalmanuscript.Competing interestsThe authors declare that they have no competing interests.Received: 6 December 2011 Accepted: 4 April 2012Published: 4 April 2012References1. Pope CA III, Burnett RT, Thun MJ, Calle EE, Krewski D, Ito K, Thurston GD:Lung Cancer, Cardiopulmonary Mortality, and Long-term Exposure toFine Particulate Air Pollution. JAMA 2002, 287:1132-1141.2. Laden F, Schwartz J, Speizer FE, Dockery DW: Reduction in fine particulateair pollution and mortality: Extended follow-up of the Harvard Six Citiesstudy. Am J Respir Crit Care Med 2006, 173(6):667-672.3. Beelen R, Hoek G, van den Brandt PA, Goldbohm RA, Fischer P,Schouten LJ, Jerrett M, Hughes E, Armstrong B, Brunekreef B: Long-termeffects of traffic-related air pollution on mortality in a Dutch cohort(NLCS-AIR study). Environ Health Perspect 2008, 116(2):196-202.4. Nafstad P, Haheim LL, Oftedal B, Gram F, Holme I, Hjermann I, Leren P:Lung cancer and air pollution: a 27 year follow up of 16 209 Norwegianmen. Thorax 2003, 58(12):1071-1076.5. Raaschou-Nielsen O, Andersen Z, Hvidberg M, Jensen SS, Ketzel M,Sorensen M, Loft S, Overvad K, Tjonneland A: Lung Cancer Incidence andLong-Term Exposure to Air Pollution from Traffic. Environ Health Perspect2011, doi: 10.1289/ehp.1002353.6. Katanoda K, Sobue T, Satoh H, Tajima K, Suzuki T, Nakatsuka H, Takezaki T,Nakayama T, Nitta H, Tanabe K: An Association Between Long-TermExposure to Ambient Air Pollution and Mortality From Lung Cancer andRespiratory Diseases in Japan. J Epidemiology 2011, 21(2):132-143.7. Statistics Canada: Mobility and Migration. 2007, http://www12.statcan.ca/census-recensement/2006/rt-td/mm-eng.cfm [accessed 25 August 2011].8. Urayama KY, Von Behren J, Reynolds P, Hertz A, Does M, Buffler PA: Factorsassociated with residential mobility in children with leukemia:implications for assigning exposures. Ann Epidemiol 2009, 19(11):834-840.9. Hurley SE, Reynolds P, Goldberg DE, Hertz A, Anton-Culver H, Bernstein L,Deapen D, Peel D, Pinder R, Ross RK: Residential mobility in the CaliforniaTeachers Study: implications for geographic differences in disease rates.Soc Sci Med 2005, 60(7):1547-1555.10. Kan K: Residential mobility and social capital. J Urban Econ 2007,61(3):436-457.11. Oishi S: The Psychology of Residential Mobility. Perspect Psychol Sci 2010,5(1):5.12. Meliker JR, Sloan CD: Spatio-temporal epidemiology: Principles andopportunities. Spatial and Spatio-temporal Epidemiol 2011, 2(1):1-9.13. Gallagher LG, Webster TF, Aschengrau A, Vieira VM: Using residentialhistory and groundwater modeling to examine drinking water exposureand breast cancer. Environ Health Perspect 2010, 118(6):749.14. Behren JV, Reynolds P, Gunier RB, Rull RP, Hertz A, Urayama KY, Kronish D,Buffler PA: Residential traffic density and childhood leukemia risk. CancerEpidemiol Biomarkers & Prev 2008, 17(9):2298.15. Boscoe FP: The Use of Residential History in Environmental Health StudiesGeospatial Analysis of Environmental: Health; 2011, 93-110.16. Dockery DW, Pope CA, Xu X, Spengler JD, Ware JH, Fay ME, Ferris BG Jr,Speizer FE: An association between air pollution and mortality in six U.S.cities. N Engl J Med 1993, 329(24):1753-1759.17. Beeson WL, Abbey DE, Knutsen SF: Long-term concentrations of ambientair pollutants and incident lung cancer in California adults: results fromthe AHSMOG study. Adventist Health Study on Smog. Environ HealthPerspect 1998, 106(12):813.18. Bellander T, Berglind N, Gustavsson P, Jonson T, Nyberg F, Pershagen G,Jarup L: Using Geographic Information Systems to Assess IndividualHistorical Exposure to Air Pollution from Traffic and House Heating inStockholm. Environ Health Perspect 2001, 109(6):633-639.19. Vineis P, Hoek G, Krzyzanowski M, Vigna-Taglianti F, Veglia F, Airoldi L,Autrup H, Dunning A, Garte S, Hainaut P, Malaveille C, Matullo G,Overvad K, Raaschou-Nielsen O, Clavel-Chapelon F, Linseisen J, Boeing H,Trichopoulou A, Palli D, Peluso M, Krogh V, Tumino R, Panico S, Bueno-De-Mesquita HB, Peeters PH, Lund EE, Gonzalez CA, Martinez C, Dorronsoro M,Barricarte A, Cirera L, Quiros JR, Berglund G, Forsberg B, Day NE, Key TJ,Saracci R, Kaaks R, Riboli E: Air pollution and risk of lung cancer in aprospective study in Europe. Int J Cancer 2006, 119(1):169-174.20. Beelen R, Hoek G, van den Brandt PA, Goldbohm RA, Fischer P,Schouten LJ, Armstrong B, Brunekreef B: Long-term exposure to traffic-related air pollution and lung cancer risk. Epidemiology 2008, 19(5):702.21. Beelen R, Hoek G, Fischer P, Brandt PA, Brunekreef B: Estimated long-termoutdoor air pollution concentrations in a cohort study. Atmos Environ2007, 41(7):1343-1358.22. Hart JE, Yanosky JD, Puett RC, Ryan L, Dockery DW, Smith TJ, Garshick E,Laden F: Spatial Modeling of PM10 and NO2 in the Continental UnitedStates, 1985-2000. Environ Health Perspect 2009, 117(11):1690-1696.23. Johnson KC, Mao Y, Argo J, Dubois S, Semenciw R, Lava J: The NationalEnhanced Cancer Surveillance System: a case-control approach toHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 12 of 13environment-related cancer surveillance in Canada. Environmetrics 1998,9(5):495-504.24. van Donkelaar A, Martin RV, Brauer M, Kahn R, Levy R, Verduzco C,Villeneuve PJ: Global Estimates of Ambient Fine Particulate MatterConcentrations from Satellite-based Aerosol Optical Depth:Development and Application. Environ Health Perspect 2010,118(6):847-855, doi: 10.1289/ehp.0901623.25. Lamsal L, Martin R, van Donkelaar A, Steinbacher M, Celarier E, Bucsela E,Dunlea E, Pinto J: Ground-level nitrogen dioxide concentrations inferredfrom the satellite-borne Ozone Monitoring Instrument. J Geophys Res2008, 113:D16308.26. Environment Canada: Canadian Regional and Hemispheric O3 and NOxSystem (CHRONOS)., http://www.ec.gc.ca/air/default.asp?lang=En&xml=1A23C1F9-A464-4041-B889-79C7DE4E847E [accessed 02 June2011].27. Lall R, Kendall M, Ito K, Thurston GD: Estimation of historical annual PM2.5 exposures for health effects assessment. Atmos Enviro 2004,38(31):5217-5226.28. ESRI: Inverse Distance Weighting Manual.[http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//003000000007000000.htm], [accessed05 March 2012].29. Zhou Y, Levy JI: Factors influencing the spatial extent of mobile sourceair pollution impacts: a meta-analysis. BMC Publ Health 2007, 7:89.30. Kahn ME: New evidence on trends in vehicle emissions. RAND J Econ1996, 27(1):183-196.31. Sawyer RF, Harley RA, Cadle S, Norbeck J, Slott R, Bravo H: Mobile sourcescritical review: 1998 NARSTO assessment. Atmos Environ 2000, 34(12-14):2161-2181.32. Environment Canada: National Emission Trends for Key Air Pollutants.[http://www.ec.gc.ca/inrp-npri/default.asp?lang=en&n = 0EC58C98-],[accessed 12 October 2011].33. U.S. Environmental Protection Agency: Reactive Nitrogen in the UnitedStates: An Analysis of Inputs, Flows, Consequences, and ManagementOptions.[http://yosemite.epa.gov/sab/sabproduct.nsf/WebBOARD/INCFullReport/$File/Final%20INC%20Report_8_19_11%28without%20signatures%29.pdf], [accessed 17 October 2011].34. Argo J: Chronic disease and early exposure to air-borne mixtures: 1. Theenvironmental quality database. Environ Sci Technol 2007,41(20):7178-7184.35. Argo J: Retrospective exposure assessment with emission inventories: anew approach to an old problem. Environmetrics 1998, 9(5):505-518.36. Kasim K, Levallois P, Johnson KC, Abdous B, Auger P: Chlorinationdisinfection by-products in drinking water and the risk of adultleukemia in Canada. Am J Epidemiol 2006, 163(2):116.37. Johnson KC, Pan S, Fry R, Mao Y, The Canadian Cancer RegistriesEpidemiology Research Group: Residential proximity to industrial plantsand non-Hodgkin lymphoma. Epidemiology 2003, 14(6):687-693.38. Ramis R, Vidal E, Garcia-Perez J, Lope V, Aragones N, Perez-Gomez B,Pollan M, Lopez-Abente G: Study of non-Hodgkin’s lymphoma mortalityassociated with industrial pollution in Spain, using Poisson models. BMCPubl Health 2009, 9:26.39. Aylin P, Maheswaran R, Wakefield J, Cockings S, Jarup L, Arnold R,Wheeler G, Elliott P: A national facility for small area disease mappingand rapid initial assessment of apparent disease clusters around a pointsource: the UK Small Area Health Statistics Unit. J Publ Health 1999,21(3):289.doi:10.1186/1476-069X-11-22Cite this article as: Hystad et al.: Spatiotemporal air pollution exposureassessment for a Canadian population-based lung cancer case-controlstudy. Environmental Health 2012 11:22.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/submitHystad et al. Environmental Health 2012, 11:22http://www.ehjournal.net/content/11/1/22Page 13 of 13

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.52383.1-0220554/manifest

Comment

Related Items