Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Childhood asthma and allergies in birth cohort studies : tools for environmental exposure assessment Sbihi, Hind 2015

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

Item Metadata

Download

Media
24-ubc_2015_september_sbihi_hind.pdf [ 4.98MB ]
Metadata
JSON: 24-1.0166377.json
JSON-LD: 24-1.0166377-ld.json
RDF/XML (Pretty): 24-1.0166377-rdf.xml
RDF/JSON: 24-1.0166377-rdf.json
Turtle: 24-1.0166377-turtle.txt
N-Triples: 24-1.0166377-rdf-ntriples.txt
Original Record: 24-1.0166377-source.json
Full Text
24-1.0166377-fulltext.txt
Citation
24-1.0166377.ris

Full Text

CHILDHOOD ASTHMA AND ALLERGIES IN BIRTH COHORT STUDIES: TOOLS FOR ENVIRONMENTAL EXPOSURE ASSESSMENT.   by  Hind Sbihi  B.Eng., Ecole Polytechnique de Montreal, 2000 M.Sc, The University of British Columbia, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Population and Public Health)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)   July 2015  © Hind Sbihi, 2015  ii   Abstract Pediatric asthma and allergies represent global health problems causing substantial disability. Epidemiological research has established a link between air pollution and exacerbation of asthma. However, the role of air pollutants in relation to atopy and on the development of asthma is unclear. This thesis examines the relationship between traffic-related air pollution and the development of atopy and asthma using two complementary Canadian birth cohorts where the impact of different exposure assessment approaches on observed associations was evaluated. Hopanes in house dust, collected in the Canadian Healthy Infant Longitudinal Development (CHILD) birth cohort study, were evaluated as markers of indoor infiltrated traffic-related air pollution by measuring their correlation with geographic predictors of outdoor concentrations of nitrogen dioxide. This correlation was dependent on the inclusion of behavioral characteristics, hindering the utility of measuring hopanes in settled dust for exposure assessment.  As an alternative approach to assess exposures in CHILD, city-specific land use regression models, questionnaires and home assessments were used to model personal exposure, including accounting for indoor/outdoor infiltration and time-activity patterns, in relation to early atopy. Spatio-temporally adjusted exposure in the first year of life was positively associated with sensitization to common food or inhalant allergens at age 1 (Odds ratio [95% confidence interval] per interquartile increase in nitrogen dioxide = 1.16 [1.00 – 1.41]).  Because atopy is often a precursor for allergic asthma, 10 years of longitudinal data from the Border Air Quality Study population-based birth cohort were used to evaluate the role of air pollution on asthma development. An interquartile range increase in nitrogen dioxide, adjusted for temporal and spatial variability, increased incident asthma among preschool (age 0-5) children by 9% (95% confidence iii  interval: 4 – 13%). Surrounding residential greenness mitigated this effect. In further analysis, the course of asthma was found to follow three trajectories: transient asthma, early-, and late-infancy chronic asthma, the latter two being significantly associated with fine particulate matter and nitrogen dioxide. This dissertation highlights the importance of integrating temporal and spatial variation in traffic-related air pollution exposure assessment and clarifies the role of early exposures on atopy and asthma initiation. iv   Preface This doctoral dissertation summarizes the objectives, methods and results of an overall research plan that was developed under the guidance of my thesis committee following the presentation and acceptance of an initial research proposal. Four research chapters (Chapters 2-5) were written as separate manuscripts intended for publication in peer-reviewed journals. These manuscripts were reviewed by numerous co-authors given that this research was made possible by a multi-university and government agencies collaboration, as well as by my doctoral committee members. For each of the chapters 2 to 5, I developed the study design, conducted all analyses, interpreted the results and prepared the manuscript. I revised manuscripts according to the feedback provided by the committee members as detailed below.  This research was approved by the University of British Columbia Behavioural Research Ethics Board (certificate number H04-80161 for the Border Air Quality Study (BAQS) cohort; project title: Establishment of a childhood respiratory disease cohort using BCLHD/Analysis of birth outcomes in the Greater Vancouver Regional District using the BC Perinatal Database; certificate number: H11-03231 for the Canadian Healthy Infant Longitudinal Development (CHILD) birth cohort; project title: Canadian Healthy Infant Longitudinal Study; as well as by the Clinical Research Ethics Board (certificate number: H07-03120 for the hopanes study within ‘CHILD’). A version of Chapter 2 has been published [Sbihi H, Brook JR, Allen RW, Curran JH, Dell S, Mandhane P, Scott JA, Sears MR, Subbarao P, Takaro TK, Turvey SE, Wheeler AJ, Brauer M. (2013). A new exposure metric for traffic-related air in settled indoor house dust. Environmental Health 12(1)]. HS formulated the research question, gathered spatial data and extracted questionnaire data, conducted the statistical analysis and led the writing of the manuscript. MB guided the study design, provided spatial data for Vancouver, gave critical input in data conditioning for the analysis and revised the manuscript for its intellectual content. RA provided the spatial data for Edmonton and Winnipeg and guided the data v  harmonizing across different spatial databases. JB gathered the ambient air samples from the national air monitoring stations and oversaw the chemical analysis of hopanes samples in the Environmental Canada Laboratory. JC performed a preliminary analysis examining personal hopanes in dust and air samples. JB, JS, TK, RA, and MB were part of the Exposure Working Group for the CHILD study where all questionnaires used in this study were designed. This group, along with the site leaders and CHILD PI oversaw the collection and initial processing of CHILD dust samples. ST, PM, PS are site leaders for the CHILD study, they are responsible for the coordination and training of research technicians and data collection. MS is the Principal Investigator of the CHILD study and has revised the manuscript drafts. SD is the Principal Investigator of the Toronto Child Health Evaluation Questionnaire (TCHEQ) study and revised the manuscript. AW has led the Windsor Ontario Exposure Assessment Study (WOEAS), participated in preparing the spatial data for Windsor, and has given critical input to the manuscript drafts. A version of Chapter 3 has been published [H.Sbihi,, R. Allen, A Becker, J Brook, J. Scott, P Mandhane, M.Sears, P. Subbarao, T. Takaro, S.Turvey, M. Brauer. 2015. Perinatal exposure to traffic-related air pollution and development of atopy in a Canadian birth cohort study. Environmental Health Perspectives. DOI:10.1289/ehp.1408700]. HS and MS formulated the research question. HS defined the specific objectives and analytical approaches to conduct the study. HS gathered environmental researcher-collected and governmental exposure data, wrote the manuscript and subsequent revisions. AB, ST, PM and PJ were site leaders for CHILD cities and oversaw the biological samples collection as well as the administration of questionnaires. TT and JB are leaders of the Environmental Working Group and have designed and facilitated access to the time-activity patterns questionnaire data. RA has provided the infiltration model and insight into using it in the context of a Canadian study. MS is the principal investigator of CHILD and has revised the manuscript drafts.  A version of Chapter 4 has been submitted for publication. HS, MB, MK defined the research theme. LT prepared administrative health data linkages provided by Population Data BC. HS and LT designed analytical and statistical methods. HS analyzed the data, interpreted the results and wrote the paper. All vi  authors discussed analyses, interpretation and presentation. All authors have contributed to and approved the manuscript. Perry Hystad is acknowledged for assigning greenness exposures to study participants. A version of Chapter 5 has been submitted for publication. HS and MB conceptualized the research objectives and formulated the research questions. HS designed the analytical approach, prepared and conducted the analysis. HS wrote the manuscript and included edits provided by MB and MK. MK provided feedback on revisions of the manuscript. Disclaimer for Chapters 4 and 5: “All inferences, opinions, and conclusions drawn in this research are those of the authors, and do not reflect the opinions or policies of the Data Steward(s).” vii   Table of Contents Abstract ......................................................................................................................................................... ii Preface ......................................................................................................................................................... iv Table of Contents ........................................................................................................................................ vii List of Tables ................................................................................................................................................ x List of Figures ............................................................................................................................................ xiii List of Abbreviations .................................................................................................................................. xv Acknowledgments ..................................................................................................................................... xvii Chapter 1: Introduction ................................................................................................................................. 1 1.1 Background and significance .................................................................................................... 1 1.1.1 Asthma and atopy in children ........................................................................................... 1 1.1.2 Risk factors for asthma and allergy ................................................................................... 6 1.1.3 Air pollution as a risk factor for asthma and allergy ....................................................... 12 1.1.4 Greenness, asthma and allergies ..................................................................................... 29 1.1.5 Summary of major gaps identified .................................................................................. 31 1.2 Data sources ............................................................................................................................ 31 1.2.1 Canadian healthy infant longitudinal development (CHILD) study ............................... 32 1.2.2 Border air quality study (BAQS) .................................................................................... 38 1.3 Objectives ............................................................................................................................... 43 1.4 Dissertation structure .............................................................................................................. 45 Chapter 2: Assessing hopanes as an exposure assessment tool for TRAP exposure .................................. 46 2.1 Introduction ............................................................................................................................. 46 2.2 Methods ................................................................................................................................... 49 2.2.1 Population ....................................................................................................................... 49 2.2.2 Hopanes ........................................................................................................................... 50 2.2.3 Geographic predictor variables ....................................................................................... 52 2.2.4 Questionnaires ................................................................................................................. 54 2.3 Statistical analysis ................................................................................................................... 56 2.4 Results ..................................................................................................................................... 58 2.4.1 Outdoor vs. indoor hopane concentrations comparison .................................................. 58 viii  2.4.2 Pooled results .................................................................................................................. 59 2.4.3 City-specific modeling results ........................................................................................ 60 2.5 Discussion ............................................................................................................................... 64 Chapter 3: Early-life exposure to traffic-related air pollution and development of atopy in the CHILD birth cohort study ........................................................................................................................................ 69 3.1 Introduction ............................................................................................................................. 69 3.2 Methods ................................................................................................................................... 71 3.2.1 Study population ............................................................................................................. 71 3.2.2 Skin prick tests ................................................................................................................ 71 3.2.3 Air pollution exposure assessment .................................................................................. 72 3.3 Questionnaires and home inspections ..................................................................................... 76 3.4 Statistical analysis ................................................................................................................... 77 3.5 Results ..................................................................................................................................... 79 3.5.1 Population characteristics and atopy ............................................................................... 79 3.5.2 Exposure levels and association with atopy development .............................................. 82 3.6 Discussion ............................................................................................................................... 88 Chapter 4: Asthma development in relation to air pollution and greenness in the BAQS birth cohort: a follow up from birth to age 10. ................................................................................................................... 92 4.1 Introduction ............................................................................................................................. 92 4.2 Methods ................................................................................................................................... 93 4.2.1 Cohort description ........................................................................................................... 93 4.2.2 Confounders .................................................................................................................... 94 4.2.3 Incident asthma case definition ....................................................................................... 94 4.2.4 Exposure assessment ....................................................................................................... 94 4.2.5 Analytical methods ......................................................................................................... 96 4.3 Results ..................................................................................................................................... 97 4.3.1 Asthma in preschool years ............................................................................................ 102 4.3.2 Asthma in school age years ........................................................................................... 105 4.3.3 Joint models .................................................................................................................. 106 4.3.4 Sensitivity analyses ....................................................................................................... 107 4.4 Discussion ............................................................................................................................. 111 Chapter 5: Pediatric asthma patterns and effects of environmental factors. ............................................. 115 5.1 Introduction ........................................................................................................................... 115 5.2 Methods ................................................................................................................................. 117 ix  5.2.1 Cohort description ......................................................................................................... 117 5.2.2 Asthma case definition .................................................................................................. 118 5.2.3 Exposure assessment ..................................................................................................... 118 5.2.4 Analytical methods ....................................................................................................... 119 5.3 Results ................................................................................................................................... 120 5.3.1 Cohort and asthma trajectories description ................................................................... 120 5.3.2 Impact of environmental factors on asthma trajectories ............................................... 127 5.3.3 Co-exposures models .................................................................................................... 131 5.4 Discussion ............................................................................................................................. 133 Chapter 6: Discussion and conclusions ..................................................................................................... 137 6.1 Summary and contributions .................................................................................................. 138 6.1.1 Key findings .................................................................................................................. 138 6.1.2 Synthesis ....................................................................................................................... 143 6.1.3 Strengths and limitations ............................................................................................... 147 6.2 Recommendations for future research .................................................................................. 153 6.2.1 Future directions for exposure assessment research ..................................................... 154 6.2.2 Future directions for asthma and allergies research ...................................................... 156 Bibliography ............................................................................................................................................. 159 Appendices ................................................................................................................................................ 181    x   List of Tables Table 1.1: Inclusion and exclusion criteria for participants’ recruitment in the CHILD study .................. 32 Table 1.2: CHILD study schedule from recruitment to age 5 ..................................................................... 34 Table 1.3: Principal exposure assessment methods used in the CHILD Study to characterize the physical environment during pregnancy and infancy periods ................................................................................... 36 Table 1.4: LUR models for nitrogen dioxide (variables , estimated annual means for all participants and standard deviation (sd)). .............................................................................................................................. 37 Table 2.1: Harmonized GIS variables across TCHEQ, WOAES, and CHILD cities ................................. 53 Table 2.2: City-specific GIS variables and buffer sizes extracted from LUR model surfaces. .................. 54 Table 2.3: Descriptive summary of questions found (as shown with a check mark) in the questionnaires delivered during home visits, recoded for analysis in the pooled investigation of hopanes in dust and land use determinants of traffic pollution. .......................................................................................................... 55 Table 2.4: City-specific determinants of hopane concentrations in house-settled dust. ............................. 61 Table 2.5: Geometric Mean (GM) and Geometric Standard Deviation (GSD) of total hopanes concentrations in the study population, by room, by homes and correlation with city-specific modeled NO2 ............................................................................................................................................................. 63 Table 3.1: Particle Infiltration (Finf) variables; mapping questions used in the MESA-Air (Allen et al. 2012) and CHILD cohorts. ......................................................................................................................... 76 Table 3.2: Atopic outcomes (inhalant, food and any allergies) by city among 2477 CHILD participants with valid skin prick test at age 1 ................................................................................................................ 79 Table 3.3: Proportion of positive responses to individual allergen skin prick tests by CHILD city (in (%)) .................................................................................................................................................................... 80 xi  Table 3.4: Cohort characteristicsb among 2477 children at one year of age with valid skin allergy tests, and crude Odds Ratio (OR) for sensitization to any allergens with 95% Confidence Interval (95% CI) ... 81 Table 3.5: Adjusted Odds Ratios (aOR) for risk of atopy per 10 µg/m3 increase in NO2 exposures during pregnancy and the first year of life. ............................................................................................................ 84 Table 4.1: Case and control characteristics (n (%) for categorical covariates; mean (standard deviation) for continuous covariates) for children meeting the case definition, matched controls by time window. .. 98 Table 4.2 : Cohort characteristics (n (%) for categorical covariates; mean (standard deviation) for continuous covariates) for control children and all non-asthmatics by cohort age groups ......................... 99 Table 4.3: Mean exposure levels, standard deviation (SD), and Interquartile Range (IQR) among cases and controls ............................................................................................................................................... 101 Table 4.4: Adjusted OR for new asthma onset relative to an interquartile (IQR) increase in air pollutants .................................................................................................................................................................. 103 Table 4.5: Exposure levels (N, mean) by quartile in the preschool age analysis among cases and controls .................................................................................................................................................................. 104 Table 4.6: Effects of surrounding greenness on incident asthma risk, adjusted for air pollutants in preschool and school children. .................................................................................................................. 107 Table 5.1: Summary statistics of number of elapsed years of data coverage by birth year ...................... 121 Table 5.2: Risk factor distribution (frequency counts for categorical variables; mean and standard deviation (sd)) by asthma trajectory groups among 68195 children. ........................................................ 122 Table 5.3: Posterior probability for each trajectory at each year of age (Non-Asthmtics; Transient; LICA: Late Infancy Chronic Asthma; EICA: Early Infancy Chronic Asthma) and 95% confidence intervals ... 125 Table 5.4: Odds Ratio (OR) of adhering to a trajectory group relative to risk factors based on weighted multinomial logistic regression ................................................................................................................. 126 Table 5.5: Exposure levels of greenness, IDW-derived pollutants and LUR-derived PM2.5, by trajectory among CHILD study participants ............................................................................................................. 128 xii  Table 5.6: Multinomial logistic regressions of unadjusted and adjusted  single exposure models:  greenness (NDVI),nitrogen dioxide (NO2), and fine particulate matter (PM2.5) models by asthma trajectories in BAQS children born between 1999 and 2002. ................................................................... 130 Table 5.7: Co-pollutants multinomial logistic regression of air pollution ................................................ 133  xiii   List of Figures Figure 1.1: Pathways linking natural spaces to human health (adapted from (Hartig et al. 2014) ). Putative causal pathways in blue, effect modifers in brown (personal characteristics (e.g. age, gender) and other intrinsic effect modifiers are not depicted for clarity). ............................................................................... 30 Figure 1.2: Study area of GVRD plus within the Border Air Quality Study .............................................. 39 Figure 2.1: Association between outdoor air and house dust hopanes major monomer ((17α(H), 21β(H)-Hopane) relative abundance. ....................................................................................................................... 58 Figure 3.1: Adjusted Odds Ratio for risk of atopy per 10 µg/m3 increase in NO2 exposures (triangle) temporally adjusted at birth address, (circle) temporally adjusted and accounting for residential mobility. .................................................................................................................................................................... 85 Figure 3.2: Adjusted OR of atopy per 10µg/m3 NO2 increase during the first year stratified by: (A) time-activity patterns (defined by the city-specific median hours per day based on the three questionnaires submitted after birth around 3,6, and 12 months) among children spending more time (n=976) and those spending less time (n=1026) away from the home; and (B) daycare facilities attendance among daycare attendees (n=765) and children never attending daycare (n=1236). ........................................................... 86 Figure 3.3: Adjusted OR of risk of atopy in CHILD families for a 10 µg/m3 increase in NO2 during the first year of life, stratified by presence of siblings in participants families (group with no siblings, n= 1085; group with sibling, n= 874). All models are adjusted for the same covariates as those used in the main analysis ............................................................................................................................................... 87 Figure 3.4: Adjusted Odds Ratio of risk of atopy for 10 µg/m3 increase in NO2 during the first year of life stratified by season (defined using weekly average of 18°C as cutoff to define cold and warm) and by home PM infiltration status (defined based on city-specific 80th percentile predicted household PM infiltration efficiency; ................................................................................................................................. 88 Figure 4.1: Adjusted Odds Ratio of incident asthma in children meeting the case definition in the (A) preschool and (B) school groups ............................................................................................................... 106 xiv  Figure 4.2: Effect modification of greenness in 100m buffer and air pollution estimates derived using inverse distance weighting (IDW) and/or land use regression (LUR)models on new asthma onset among preschool age children by (A) gestational length (<37 weeks) and (B) birth weight (<2500g) ............... 108 Figure 4.3: Risk of asthma incidence in the preschool group (n=6948), stratified by (A) sex and by (B) parity ......................................................................................................................................................... 109 Figure 4.4: Stratified asthma risk analysis by three indicators of Socio-economic status (A: Household Income; B: Maternal Education status; C: Maternal age at birth) ............................................................ 110 Figure 5.1: Group memberships and confidence intervals for the posterior distributions of probability of asthma at each time point between birth and age 10 for each of the trajectories in BAQS cohort ........... 124 Figure 5.3: Scatter plot matrix of air pollutants and greenness exposure by asthma trajectory and exposure distributions among children with one of the three asthma trajectories. ................................................... 132  xv   List of Abbreviations BAQS: Border Air Quality Study CO: Carbon monoxide CHILD: Canadian Healthy Infant Longitudinal Development Der p : Dermatophagoides pteronyssinus  Der f: Dermatophagoides farinae Finf: Infiltration efficiency GIS: Geographic Information System GVRD: Greater Vancouver Regional District IDW: Inverse-Distance Weighting  IgE:  Immunoglobulin E LUR: Land Use Regression NDVI: Normalized Difference Vegetation Index  NO: Nitrogen monoxide NO2: Nitrogen Dioxide O3: Ozone PM10: Particulate matter with an aerodynamic diameter <10um PM2.5: Particulate matter with an aerodynamic diameter <2.5um  xvi  SO2: Sulfur dioxide TRAP: Traffic-Related Air Pollution TCHEQ: Toronto Child Health Evaluation Questionnaire WOEAS: Windsor Ontario Exposure Assessment Study xvii  Acknowledgments  It is with extreme pleasure that I write this section where I have the opportunity to express my gratitude to all who have made this journey possible and often enjoyable. First and foremost, Dr. Michael Brauer, my supervisor, has not only been a very supportive, challenging, and insightful supervisor but also a mentor who taught me by example to teach with respect, collaborate with integrity, and conduct research with rigor. Members of my committee, Drs. Mieke Koehoorn and Ryan Allen, provided me with great support and a lot of “food for thought” following our one-on-one meetings throughout the PhD. I am also grateful for the opportunities that Dr. Allen had offered outside of the PhD research work that helped complete my learning experience: reviewing journal manuscripts, giving seminars, and discussing ideas that shaped the trajectory of my research. Overall, I believe I have had the best supervisory team with immense intellectual generosity, thank you! Thank you to the CHILD study team, and especially to the Exposure Working Group: Drs. Malcom Sears, Jeffrey Brook, Tim Takaro, James Scott, Stuart Turvey, as well as the many co-authors from this pan-Canadian collaborative effort who have made this work possible. Your input and suggestions are greatly appreciated. Dr. Diana Lefebvre provided me with tremendous support and helped facilitate the work with the CHILD team, thank you Diana! I would like to thank Lillian Tamburic for always carving time out for my multiple questions related to the research related to the BAQS birth cohort. Your expertise, friendship and support made this work immensely pleasurable. I have learned a lot from you.  Big thanks to my friends from the School of Population and Public Health (a special mention to my 2.5 years office mate and dear friend Dr. Tracy Kirkham, and to Annie Wang and Emily Rugel) and outside xviii  the UBC community: through our interactions and social activities you are the support network that helped me keep enjoying the graduate school experience. I would like to thank the University of British Columbia, Killam Trusts, Canadian Institutes of Health Research, and the province of British Columbia BC Clean Air Research for their generous financial support.  To my parents, you are and always will be the “gold standard”, thank you so much for your unwavering support.  To my siblings, thank you for picking up the phone in the middle of the night to catch me in the daytime here in Vancouver and to simply make sure that I was all right balancing work, school and family. To Ismaël, Amin and Ali: you are my pillars. Without your patience and love, I would not have achieved this work.  1  Chapter 1: Introduction 1.1 Background and significance 1.1.1 Asthma and atopy in children Respiratory and allergic disorders in children are a serious health issue due to their high prevalence worldwide and negative impact on the quality of life during childhood and later throughout adulthood (Bousquet et al. 2005; McColley and Morty 2012).  Traditionally, asthma has been viewed as an allergic disease as demonstrated in several studies where atopy strongly predicts its development (Custovic et al. 2013; Simpson et al. 2010). However, children can also develop asthma without being atopic. Atopy, the condition referring to allergic manifestations following exposure to a trigger allergen, is also an important health issue and can share common etiology with asthma. In light of the substantial geographic variations, both for prevalence rates and trends (Asher et al. 2006; Eder et al. 2006; Pearce et al. 2007), these conditions are likely multifaceted and include not only genetic factors but also environmental and lifestyle factors (Nicolaou et al. 2005; Subbarao et al. 2009). Air pollution constitutes an environmental exposure that has captured particular attention, especially given the results of a child-specific health impact analysis showing health and economic benefits after a reduction in air pollution levels (Wong et al. 2004). In this chapter, the clinical manifestations and characteristics of both asthma and atopy are presented, with an overview of their natural history and epidemiology. The role of known and suspected environmental exposures will be described with a focus on neighborhood exposures: air pollution, especially traffic-related pollutants, and green spaces.  1.1.1.1 Clinical presentation  Asthma is a heterogeneous disease that can be described as a chronic inflammatory disorder of the airways, characterized by intermittent airflow obstruction, shortness of breath and wheeze (Lemanske Jr. 2  and Busse 2010). The underlying pathophysiological basis of asthma is believed to be inflammatory processes involving cells of the innate and adaptive immune systems. Cytokines, T and B lymphocytes, mast cells, and histamines act together with epithelial cells to cause bronchial hyper-reactivity, a characteristic smooth muscle contraction involving mucous secretion and other changes that lead to airway remodeling if these changes remain unresolved (Maddox and Schwartz 2002). Airway remodeling is thus the irreversible end result of such pathophysiological mechanisms, also referred to as endotypes1. Clinically, children often have allergic asthma which is initiated by allergic sensitization. Following this, they may develop allergic rhinitis or other allergic conditions (e.g. eczema) and progress to asthma, a trajectory often referred to as the ‘atopic march’. However non-atopic asthma also presents similar clinical manifestations and represents a non-negligible 44% of pediatric asthma in the United States (US) (Arbes et al. 2007), a proportion that has considerable geographic variation (Ronchetti et al. 2009).  Atopic diseases constitute a group of allergic disorders (e.g.  allergic asthma, atopic dermatitis, and allergic rhino-conjunctivitis) that are subject to hereditary predisposition, called atopy. Thus atopy is a personal and/or familial propensity to developspecific immunoglobulin E (IgE) antibodies against common allergens. Sensitization is defined as the presence of a specific immune mechanism directed towards a particular antigen and is most often shown by a positive skin-prick test and/or by the presence of specific IgE (sIgE) antibodies. This definition of sensitization does not necessarily imply the presence of allergic symptoms. In fact, the more recent view is that quantification of atopy, either as the level of sIgE and/or the size of skin test wheal diameter, can help identify those at risk of persistent allergic disorders including allergic asthma (Custovic et al. 2013).                                                       1 Endotype (can also be termed endophenotype): subtype of disease defined by distinct pathophysiological mechanism. A phenotype, which does not point to an underlying pathophysiological process,  represents the observable manifestations of each individual’s genetics that can change in response to different environmental stimuli (Corren 2013)  3   While asthma is traditionally categorized into allergic versus non-allergic asthma, there is an increasing recognition that this paradigm has been an oversimplification. Different asthma phenotypes, described by clinical features, are now being clustered by one or several endotypes (Corren 2013; Henderson 2014a; Lötvall et al. 2011). Conversely, a discrete endotype, such as eosinophilic airway inflammation, can be present in several phenotypes including atopic asthma or severe recurrent wheeze whereas neutrophil-predominant airway inflammation was associated with chronic obstructive pulmonary disease in adults and with severe, non-steroid responsive childhood asthma (Henderson 2014a). The identification of all endotypes underpinning the various phenotypes of asthma have yet to be uncovered (Belgrave et al. 2013). Birth cohort studies with long-term follow up can aid in recognizing clinically similar groups of phenotypes or sub-types of asthma.  1.1.1.2 Natural history, epidemiology Asthma is the most common chronic disease in childhood and a leading cause of emergency department visits, hospitalizations and school absenteeism in children  (Bousquet et al. 2005; Ismaila et al. 2013; Masoli et al. 2004). In Canada, based on two national surveys, the prevalence among children aged 0 to 11 years who had been diagnosed with asthma was estimated to be at 11% in the mid-nineties (Millar and Hill 1998); and by 2001 the rate had risen to more than 13%, with a sharper increase among 0- to 5- year olds (Garner and Kohen 2008). Current estimates of childhood asthma prevalence in Canada have shown a plateau (12% to 13%), following an approximate four-fold increase over the past twenty five years (Subbarao et al. 2009). Similar trends have been observed in most developed countries (Eder et al. 2006). Developing countries are likely to follow the increasing trends seen in developed countries where the observed increases are likely related to environmental changes associated with the adoption of a ‘modern’ lifestyle (Nicolaou et al. 2005).  4  In the late nineties, global studies of asthma and atopy have shown increases in the prevalence of wheezing,  increased in areas where prevalence was initially low (e.g. Asia) (Asher et al. 2006). Prevalence either decreased or remained unchanged in areas where prevalence was already high (Asher et al. 2006). After the 1990s, estimates of temporal trends in the prevalence of asthma in several European and Asian countries were conflicting, possibly because of the greater variation seen in trends in the prevalence of symptoms suggestive of asthma compared to those of diagnosed asthma (Eder et al. 2006). The global variations in asthma prevalence may thus be at least partially related to differences in asthma symptoms definition. Therefore, the main challenge in pediatric asthma has been and still remains to be the characterization of this “heterogeneous condition” (Bisgaard and Bønnelykke 2010). This would in turn imply that childhood asthma does not follow a specific trajectory but that there may be several natural histories to unveil and understand.  Given the evolving and complex nature of asthma, long-term studies are necessary to understand the development of the disease, to identify associations between exposure and the initiation of the disease in order to gain more knowledge about its different phenotypes. One landmark cohort study, the Tucson Children’s Respiratory Study (Martinez et al. 1995b) allowed the first segmentation of asthma among children:  growing out of symptoms, growing into symptoms, or continuing to have symptoms. Cohort studies have shown that the majority of children belong to the first segment with only 10 to 15% of children being diagnosed with “true” asthma at the time they are at school when 50% of preschool children have wheezing symptoms (Sears et al. 2003). Since the Tucson study, asthma in early infancy and childhood have been further characterized and are now commonly described as transient wheezing, non-atopic wheezing, late-onset wheezing and persistent wheezing.  The assessment of asthma in epidemiologic studies is complex in light of the various trajectories and phenotypes. To date, there is no single instrument to identify asthma with certainty. The “gold standard” is physician-diagnosed asthma, a diagnosis made on the basis of medical history, physical examination, 5  and assessment of airway obstruction. Studies usually rely on questionnaires where participants either self-report asthma based on symptoms, or report receiving asthma diagnosis from a physician. Atopy is a consistent risk factor for asthma in many epidemiological studies (Custovic et al. 2013). The proportion of asthma attributable to atopy in children has been estimated to be 38%, but there is considerable variation between studies. The multicenter International Study of Asthma and Allergies in Childhood (ISAAC) phase II showed that the population fraction of asthma attributable to atopy differed greatly between countries according to their economic development, being 40.7% in study centres from  economically affluent countries and 20.3% in centres from ‘non-affluent’ countries (Weinmayr et al. 2007).  1.1.1.3 Sensitive Periods of Exposure Environmental exposures that occur in early-life are the most influential for respiratory and immune systems (Selevan et al. 2000). Newborns have an imbalanced immunity at birth with both an immature innate immune system and an acquired immune system skewed toward Type 2 helper cell (Th2). Adjustments in the T-helper cells and further maturation of immune cells through interaction with environmental factors such as microbial agents are needed for the infant to restore balance among Th-driven responses and to aid in the complete maturation of the innate immune system, which would minimize the risk of Th2-mediated allergic disease (Dietert and Zelikoff 2008). Early observations in farming cohorts where some aspect of farm living in early life (i.e. microbial stimulation) appeared to protect against the development of allergic diseases, and gave support to the role of immune balance efficiency in relation to the development of allergic diseases. This “hygiene hypothesis” stipulates that the altered patterns of exposure to microorganisms resulting from more western lifestyles leads to immune malfunction characterized by the imbalance between Type 1 and Type2 helper T cells [Th1] and [Th2] (Strachan 2000). Any early life environment exposures, including air pollutants, hindering these adjustments towards the restoration of Th-cells balance could be a risk factor for allergic diseases and 6  asthma. In addition, two main points were summarized in the Workshop to Identify Critical Windows of Exposures for Children’s health. The first conclusion was that interference with normal development of the respiratory and immune systems during the intrauterine period and infancy may have lasting consequences. Second to this, it was suggested that immaturity of metabolic pathways hinders the ability of fetuses and young children to clear toxicants effectively (Dietert et al. 2000).  There are many plausible reasons for the vulnerability of young children: they are exposed to higher doses since they breathe more air than adults relative to their body weight and pollutants have more impact on the rapidly developing systems of fetuses and children (Dietert et al. 2000).  The vulnerability of this segment of the population has been consistently reported in the epidemiological literature(Aguilera et al. 2013; Jedrychowski et al. 2005; Morales et al. 2015). For example, second-hand tobacco smoke exposure during early life affects asthma risk (Rosa et al. 2011), while exposure in later life is associated with allergic disorders, asthma exacerbations, but not new asthma onset (Gilliland et al. 2001). While numerous exposures are known to contribute to childhood asthma exacerbations, including ambient air pollution (Tzivian 2011), fewer factors have been well established as risk factors for the development of pediatric asthma, resulting in a long list of other putative risk/protective factors.  1.1.2 Risk factors for asthma and allergy Environmental exposures, especially those encountered during pregnancy as well as in infancy appear to play an important role in the development of pediatric asthma and allergic diseases given the development and continuing maturation of the immune system and the lungs during these periods. Recent studies, mostly birth cohort studies conducted in developed countries, have suggested that exposure to traffic-related outdoor air pollution (TRAP) is related to incident asthma while others have linked common allergens of indoor origin and endotoxins with the development of childhood asthma and wheeze.  7  The understanding of the contribution made by environmental exposures to asthma and allergies development in children requires the assessment of exposures in all micro-environments where the child develops: in utero, indoors and outdoors. Pregnant women, infants and young children are exposed to a wide range of environmental factors which include those that can be inhaled (indoor and outdoor air), absorbed dermally, ingested, and the microbial milieu (gastro-intestinal flora for example). The main risk factors, excluding host factors, which have been known from long-term studies to be associated with asthma and allergic disorders, are genetics, behavioral (e.g. diet), and environmental (e.g. second-hand tobacco smoke, virus, and bacteria) factors. Following a brief review of these known factors (genetics, second-hand tobacco smoke (SHS), microbial agents), the role of TRAP, a suspected environmental factor for which the epidemiologic evidence is mounting, will be critically reviewed (see sub-section 1.1.3).  1.1.2.1 Genetics  Family and twin studies indicate that genetics play an important role in the study of asthma etiology (Los et al., 2001). Linkage, association studies and genome-wide screening suggest that many genes are involved in the pathogenesis of asthma. It is likely that genetics contributes to the different trajectories of asthma development. However, the dramatic and rapid increase in asthma prevalence in developed countries observed during the second half of the 20th century  (Eder et al. 2006) cannot be explained by genetics alone. Short-term changes in the occurrence of asthma are more likely to be influenced by changes in the environment, diet, and lifestyle than by changes in the genetic pool. The role of environmental influences in asthma expression is further supported by natural epidemiological observations within genetically similar populations after the German reunification (Subbarao et al. 2009). Similarly, the clear differences in asthma prevalence in populations with the same ethnic background but differing environmental exposures are convincing arguments that environmental influences cannot be ignored (Wong and Chow 2008). Thus, asthma can be labeled as a genetically complex disease where 8  multiple genetic variants interact with environmental factors to modify both susceptibility and severity of the disease.  The effects of gene-by-environment interactions in asthma are complex because interactions are affected not only by multiple gene variants and potential exposures, but also by time windows. For instance, in early childhood, genes code for enzymes that detoxify inhaled agents (e.g. glutathione transferase) while in utero genes may have a more direct effect via epigenetic mechanisms (e.g. possibility that maternal smoking modifies fetal genetic expression). However, gene-environment interaction studies for asthma have clearly demonstrated interaction effects for some environmental inhalable exposures including endotoxin (Martinez 2007; von Mutius 2009), SHS (Dizier et al. 2007; Wu et al. 2007), and air pollution which encompasses ozone exposure (London 2007; MacIntyre et al. 2014), nitrogen dioxide (MacIntyre et al. 2014; Melén et al. 2008) and particulate matter exposures (Kerkhof et al. 2010). Therefore genetics can be useful to disentangle environmental influences in specific subgroups of the population. Indeed, compared with population-based studies in which the effect of an environmental exposure is compared between groups (exposed versus unexposed), genotyping allows the identification of susceptible individuals that in turn allows for a better estimation of the true magnitude of effect of an environmental exposure for the population at risk. In sum, the role of genetics adds more specificity to the causal relationship under investigation. However, as long as the assessment of environmental exposures lacks accuracy and precision, the ability to understand the effects of gene by environment interactions will be limited.  1.1.2.2 Second Hand Smoke (SHS) SHS is likely the most important indoor pollutant and exposure of public health relevance. Tobacco smoke contains more than 4000 chemical substances which have known or suspected carcinogenic, mutagenic, toxic or inflammatory properties (California, EPA 1997). For infants and children, exposure to SHS due to maternal smoking during pregnancy is of particular importance. Carlsen and Carlsen provided 9  a review of the role of SHS and respiratory disorders in infants and young children highlighting the role of genetic polymorphisms (Carlsen and Carlsen 2008). Therefore, it appears that asthma and SHS are impacted by time of exposure, genetic susceptibility and heritability (i.e. history of atopy). Vork and colleagues performed a meta-regression to identify biases and inconsistencies among all the previous reviews and found, based on thirty eight recent studies and previous meta-analyses, that the relative risk of secondhand smoke and ever, current and incident asthma were 1.48 (95% CI 1.32 – 1.65), 1.25 (1.21 – 1-30) and 1.21 (1.03 – 1.36), respectively (Vork et al. 2007).  The weight of evidence is increasingly suggestive of a causal pathway between exposure to SHS mixtures and development of asthma under a set of genetic predispositions. 1.1.2.3   Microbial agents  The list of inhalable indoor factors, reviewed here, that have been linked to asthma development in children includes endotoxins, (1,3)β -D-Glucans (Dales et al. 2006), different indoor aero-allergens such as cockroach allergens (Salam et al. 2004), mouse allergens (Matsui et al. 2006), and house dust mites (Huss et al. 2001; Loan et al. 2003; Peat et al. 1996; Sidenius et al. 2002). The latter exposure is probably the one that has gathered the most attention from researchers in terms of exposure assessment, quantification, analysis and prevention (Loo et al. 2010).  Damp environments, Endotoxin and Beta-Glucans The consistent reporting of several epidemiological findings on the adverse health effects of living in damp homes or those with visible molds has been recently summarize in a meta-analysis by Fisk and colleagues. The associated percent increases in health outcomes (asthma, wheeze, rhinitis) for participants residing in damps homes ranged from 30% to 52%   (Fisk et al. 2007). Increased dampness stimulates growth of fungi, bacteria, and other biological agents.  10  Endotoxin, a component of gram-negative bacteria cell walls, is commonly used as a surrogate for microbial exposure. Other important bio-active components of damp environments are the (1,3)-β-glucans, which are also cell-wall components, but in this case resulting from fungal growth. Both components have inflammatory properties including immune-stimulating characteristics and airways irritation.  In contrast to the findings reported by Fisk and colleagues on visible mold and measured specific mold species, cohort and birth cohort studies showed that early-infancy exposure to (1,3)-ß-D-glucan decreases the risk of developing wheezing symptoms and parental-reported, doctor-diagnosed asthma in young children (Douwes et al. 2006; Iossifova et al. 2007). This finding is in line with the hygiene hypothesis. While this hypothesis involves the role of both environmental and host factors (e.g. via the lack of exposure to infections such as the respiratory syncytial virus early in life), restricting exposures to inhalable factors would postulate that the rise in allergic diseases and asthma is due to increasingly hygienic environments that do not provide sufficient microbial stimuli to allow for the normal maturation of the immune system and development of appropriate immunologic responses. This hypothesis has been extensively debated (Platts-Mills et al. 2001; von Mutius 2001). Since the original theory offered by Strachan (Strachan 2000) mainly holds for allergic asthma but does not characterize observations for non-atopic asthmatics, the hygiene hypothesis has been revisited to include more objective measures of bacterial and fungal exposure. For example, endotoxin has two opposing effects: it protects against atopy but is a risk factor for non-atopic asthma and wheezing. Recently published data tried to improve on our understanding of the impact of microbial exposure on asthma initiation by identifying microorganisms involved in the onset of the disease using two cohorts that relied on complementary techniques, namely the use of single-strand conformation polymorphism (SSCP) analyses and the culture techniques (Ege et al. 2011). These authors found a greater diversity of exposures in farms/rural areas as compared to urban settings; this diversity explained a wide proportion of the inverse relationship to asthma in the study 11  subjects. Interestingly the diversity did not show associations with atopy, as the latter was inversely associated with the level of gram-negative bacteria exposure. The contrast in findings for asthma and atopy likely indicate that there are different mechanisms by which microbial exposures affect asthma and atopy. This in turns fits the differences in genetic determinants of these two related conditions as shown in the latest Genome-Wide Association Study (GWAS) on asthma and allergies (Moffatt et al. 2010). It has been suspected that (1,3)-ß-D-glucan may have a similar impact on the development of an infant's immune system as an early endotoxin exposure (Iossifova et al. 2007; Schram-Bijkerk et al. 2006).  House Dust Mites Since Sporik’s landmark paper in 1990, in which increased exposure to house dust mites (HDM) was proposed as a risk factor for asthma (Sporik et al. 1990), the role of HDM in the development of childhood asthma has been the subject of intensive investigation (see for example (Platts-Mills et al. 2009; von Hertzen and Haahtela 2009)). Nevertheless, recent reviews demonstrate a consensus that early-life exposure to mite-related allergens increases the risk of sensitization, thereby leading to asthma and allergic disease exacerbations (e.g., eczema, rhinitis) among those sensitized (Arshad 2010; Platts-Mills et al. 2007). The question of the role of HDM and/or HDM allergens in the development of asthma remains elusive, although there is mounting evidence for a causative pathway among genetically predisposed subgroups (Bisgaard et al. 2008). Exposure to house dust mites has likely increased, especially in light of the many recent changes in the human environment that contribute to the increase of “mite load” indoors: increase use of indoor heating and air humidification systems provides optimal environmental condition for mite growth; new “green” detergents free from acaricides does not affect mites; and vacuum cleaners have replaced picking up and beating carpets outside the home leading to re-suspension of the particles carrying the HDM allergens (Heinrich 2010). Despite varying levels across the world, the link between HDM allergens and asthma 12  development in children has been shown among different populations (general population versus “at risk”, varying SES, etc.), which would argue in favor of a causal relationship. Of particular interest is the finding that suggests that the dose-response relationship may not be linear as previously reported (Celedon et al. 2007) but rather bell-shaped. The link between mite allergen and the induction of allergic sensitization has been established on the one hand, and the causal relation between specific allergic sensitization to mites and asthma onset is unequivocal on the other hand. Yet, the direct relation between mite exposure and asthma is still unclear, with both positive association (Celedon et al. 2007; Torrent et al. 2006; Tovey et al. 2008) and no effect reported in birth cohorts assessing exposure at birth (Lau et al. 2005). This would indicate that HDM allergens are associated with certain phenotypes of asthma (e.g. allergic asthma). In the most recent review on this topic, Heinrich also proposes that while no direct evidence for mite exposure and asthma onset can be drawn, “sufficient evidence” can be concluded from an indirect link via the increased risk of sensitization (Heinrich 2010).  1.1.3 Air pollution as a risk factor for asthma and allergy 1.1.3.1 Effect of ambient air pollution and respiratory diseases in children  The outdoor environment may impact respiratory health both directly and indirectly by infiltration of outdoor-generated pollutants into indoor environments. For the former association, there is sufficient evidence to suggest that air pollutants, such as ozone and particulate matter, decrease lung function, trigger asthma exacerbation, and increase rates of hospitalization for asthma. This has been observed in various locations, including Toronto, Ontario (Thurston et al. 1997) , Seattle, Washington (Schwartz et al. 1993), and many European cities (Sunyer et al. 1997) .  While studies have demonstrated that exposure to outdoor air pollutants exacerbates pre-existing asthma, the evidence for its effect on the initial development of asthma and allergic disorders remains unclear 13  (Health Effects Institute 2010; Heinrich and Wichmann 2004). The lack of conclusive evidence is possibly due to poor or inadequate exposure assessment methods. Indeed, assessing long-term exposure to air pollution has always been a significant challenge. The cruder assessment involves estimating pollution at the community level based on fixed-location government monitoring data, thereby assigning each person in the community the same level of exposure. However, within-community differences in air pollution are often as large as or larger than between-community differences. Therefore, people living in the same community may have considerably different exposures and assigning them the same exposure leads to significant misclassification. This type of error in exposure assessment, called Berkson error (Armstrong 1998), reduces the power of a study and implies that a “true” underlying association is less likely to be detected. Furthermore the threat of confounding with this approach can further obscure any associations. Indeed, community differences other than air pollution levels such as socioeconomic status may also confound any potential association between air pollution and childhood asthma and allergic disorders development. Research interest has therefore shifted to assessing within-community differences in exposure. In particular, this interest was directed towards urban within-community air pollution since several studies found that asthma rates are higher in urban areas than in rural areas, suggesting the urban environment may contribute to asthma development (Nicolaou et al. 2005). Another motivating aspect for this accrued interest is that within-city variability may also reflect different pollution sources and pollutant mixtures, often related to traffic. This is somewhat different than the between-city analyses where pollution sources encompass industrial and traffic sources, as well as secondary formation of air pollutants.  An example of a simple approach to examine urban within-community air pollution variation is an assessment based on proximity to emission sources such as major roads and highways. Indeed, proximity to traffic has drawn significant attention. Interest in investigating traffic proximity in relation to asthma development was also motivated by studies of natural experiments such as the East-West German 14  comparison. Following the reunification of Germany, it was possible to compare genetically similar populations where the main difference of exposure with respect to air pollution was that Western Germany had higher NO2 levels compared to high SO2 levels in Eastern Germany driven by coal combustion (supposedly the main heating source). Since the rates of asthma and atopy in areas of East Germany were found to be much lower than in West Germany, the focus of research has shifted from background air pollution toward pollution from motor vehicles (von Mutius et al. 1994). The literature on pollution from motor vehicles showed that focusing the exposure assessment to the urban pollution portion due to traffic emissions unveiled associations with childhood asthma development that were unseen when estimating general urban/background air pollution. At the ecological level, the ISAAC analysis phase I assessed the urban background concentration of PM10 in 55 countries in order to evaluate the role of ambient particulate matter in explaining geographical variation in the prevalence of asthma (Anderson et al. 2010). Anderson and colleagues concluded that urban background PM10 has little or no association with childhood asthma. On the other hand, using self-reported truck traffic intensity in the same study did show an association with asthma, giving further weight to shifting the focus to TRAP (Brunekreef et al. 2009). Since these studies, there is growing evidence that air pollutants that are specifically associated with traffic exposure may have a role in incident asthma (Guarnieri and Balmes 2014). A recent review concluded that living close to busy roads is an independent risk factor for the onset of childhood asthma even though it did not reach the level of sufficient evidence (Gowers et al. 2012). 1.1.3.2 Traffic-related air pollution, asthma and allergies: mechanisms, exposure assessment, and epidemiological evidence  The emission of air pollution from motor vehicles affects the quality of the air, especially considering the rapid growth of the motor-vehicle fleet worldwide and despite substantial improvements in emission controls. TRAP is a mixture of particulate matter (PM), volatile organic compounds (VOCs), polycyclic 15  aromatic hydrocarbons (PAHs), and gases such as nitrogen oxides (NO, NO2) and carbon monoxide (CO). This mixture’s composition is a function of the type of vehicle, its operating and maintenance condition, type and quality of fuel, wear of tires and brakes, and engine lubricants. Nitrogen oxides are often used as surrogates for TRAP as data for these air pollutants, associated with a diverse set of health effects, is often routinely collected by governments.  Nitrogen oxides are a group of reactive gases, of which nitrogen dioxide (NO2) and nitric oxide (NO) are most commonly monitored. NO and NO2 are both released from combustion. NO is a primary combustion pollutant that reacts with ozone in the atmosphere to form NO2. Concentrations of both gases are highest near traffic sources. Long-term average NO2 levels in cities worldwide vary from 15-60 µg/m3 (WHO 2006). In general, traffic-related pollutants impact ambient air quality on a broad spatial scale ranging from roadside, to urban, to regional background. Based on synthesizing evidence gathered by the panel of Health Effects Institute (HEI) experts, 300 to 500 meters from a major road was identified as the near-source area most impacted by traffic; variations exist depending on meteorology, background pollution, and local factors (Health Effects Institute 2010).  Particulate Matter (PM) refers to a complex and dynamic mixture that can be characterized by mass or number concentrations in specific size categories and/or by chemical composition. Monitoring efforts target particles with an aerodynamic diameter less than 10 micrometers (referred to as PM10) because they can be inhaled and deposited into the lungs. A subgroup of PM10 is fine particulate matter, or particles with an aerodynamic diameter less than 2.5 micrometers (referred to as PM2.5), which are able to penetrate deep into the lungs. For this reason, they have become increasingly monitored. PM10 includes coarse PM which, while potentially related to TRAP, might arise from non-combustion sources (e.g. brake and tire wear, suspended road dust into air by wind and traffic) as well as industrial and mobile sources. Primary sources of PM2.5 are mainly from combustion. Particulate matter levels can be higher 16  near emission sources, such as roadways (Health Effects Institute 2010), in particular for black carbon, ultrafine particles and PM10 (Karner et al. 2010). As traffic is currently, and is likely to remain, a major contributor to air pollution in urban settings, understanding its impacts on health is of upmost importance. According to the Health Effects Institute review on TRAP, there is sufficient evidence to infer a causal role for TRAP in the exacerbation of asthma in children (Health Effects Institute 2010) . However, there is also mounting evidence that TRAP may induce asthma onset as discussed below. Biological mechanisms and role of TRAP in the incidence of asthma and allergic disorders.  Experimental studies have identified specific TRAP-induced biological effects. The effect of diesel exhaust particles has received the most attention. It has been shown to enhance allergic inflammation by skewing the immune response towards Th2 activation, therefore increasing the production of IgE. D'Amato and colleagues provide a thorough review of the potential effects of exposure to ozone, NO2, SO2, and inhalable particulate matter on respiratory allergic diseases (D’Amato et al. 2002). These studies, however, argue for the allergic asthma phenotypes. Other mechanisms involving the oxidative stress pathway have also been suggested for the role of TRAP in childhood asthma development. Reactive oxygen species (ROS) and free radicals are generated by traffic related pollutants such as NO2 and PM. Oxidative stress arises when the neutralizing antioxidant defense in the airways, secreted to counteract the effects of ROS, is overwhelmed. In turn, this enhances inflammation, specifically around the respiratory epithelium, and further ROS formation (Bråbäck and Forsberg 2009). In the developing lung, air pollutants may interact with the airway in a way that modifies its structural functions leading thus to asthma (airway remodeling). More recently, epigenetic studies are increasingly supporting the oxidative stress pathway in the association between TRAP and asthma. In particular, following short-term exposure to diesel, changes in the DNA methylation was seen in gene sites involved with inflammation and 17  oxidative stress responses. This provides a plausibile pathway by which TRAP can impact gene expression and subsequently respiratory system function (Jiang et al. 2014).  The expert panel of the Health Effects Institute found evidence suggesting that TRAP was associated with pediatric asthma onset, a finding further corroborated by recent publications (Clark et al. 2010b; Gehring et al. 2010a; Gruzieva et al. 2013; McConnell et al. 2010). The most relevant studies to assess pediatric asthma onset are birth cohort studies as this design allows the follow up of study participants from birth or earlier during pregnancy, and to characterize their exposures prior to the onset of the disease, establishing the direction of causality. However, assessing the role of TRAP on childhood asthma in epidemiologic studies is fraught with issues related to the case definition. Asthma onset relies on history of symptoms that may indicate not only onset but also exacerbation. Additionally, pediatric asthma diagnosis is challenging and may often confuse wheezing with “true” asthma, since robust diagnosis is often only possible at school age. Among the first studies to highlight the role of timing of exposure for asthma onset was a case-control study in France in children aged 4 to 14 years. The authors of this study demonstrated an association between traffic density (>30.3 (veh/day)/m) before the age of 3 years and asthma (Zmirou et al. 2004) but no association with lifelong exposure,  suggesting that early exposure to traffic related air pollution plays a role in the initiation of asthma. Adding to the weight of evidence are the generally consistent results from birth cohort studies over different geographical areas spanning Asia, Europe and North America. In Europe, a Dutch cohort (the Prevention and Incidence of Asthma and Mite Allergy [PIAMA]) (Brauer et al. 2002, 2008; Gehring et al. 2010a) and two German cohorts (GINA and LISA) used a similar modeling-based exposure assessment approach to estimate individual estimates at each subject’s home address at birth with rigorous control of other covariates related to asthma and allergies. While the German studies suffered from a less robust 18  case definition, using “obstructive/asthmoid bronchitis” that likely related to the mixed results obtained in these cohorts (Gehring et al. 2002; Morgenstern et al. 2007), the Dutch study examining the children at three time points (2, 4, and 8 years of age) had more consistent results for asthma and allergies. Of particular interest is the follow up study on 4000 Dutch children until age 4 where Brauer et al. showed that exposure to higher levels of TRAP was associated with increased risks of doctor diagnosed asthma: 1.3 times increased risk per 0.6 x 10-5/m increase in level of soot, 1.2 times increased risk per 10 µg/m3 increase in NO2, and 1.2 increased risk per 3.3 µg/m3 increase in PM2.5 (Brauer et al. 2008). The follow up study led by Gehring and colleagues confirmed the association of traffic air pollutants in 8 year old children (Gehring et al. 2010a). A Canadian birth cohort study, also relying on similar modeling method to assess exposure in utero and during the first year of life showed positive associations between early-life TRAP exposures and asthma. Here, the researchers determined that exposure to traffic-derived pollutants in early years, including NO, NO2, CO, and black carbon (BC), were associated with the highest risk estimates compared to other pollutants (SO2, and PMs), although these are not good TRAP proxies  (Clark et al. 2010b). In the US, different approaches for exposure assessment to TRAP were used to assess its impact on pediatric asthma incidence. In the Cincinnati birth cohort, residing close to stop-and-go traffic (<100 m) had a 3-fold increased risk of  wheeze in the first year of life compared to unexposed infants (Ryan et al. 2005), and increasing modeled levels of traffic soot showed a dose-response relationship with infant wheezing (Ryan et al. 2007). Both studies did not have robust case ascertainment as the identified phenotypes were outcome measures that often resolve subsequently at school age.  However, the classification of the evidence for the role of TRAP on incidence of asthma by the Health Effects Institute expert panel was deemed as suggestive only. Indeed, even when including various phenotypes of asthma, some cohort studies did not find any association between TRAP and childhood asthma. For instance, a cohort of school children in Oslo, Norway where individual exposures to TRAP pollutants (NO2, PM10 and PM2.5) in early life and during lifetime were calculated using an atmospheric 19  dispersion model with contribution from busy roads showed no association with asthma onset (Oftedal et al. 2009). Similarly, negative reports from German cohorts showed no relation between exposure to air pollutants and new onset of asthma (GINI/LISA) (Gehring et al. 2002). Therefore, results to date are diverging. It should be noted that the most consistent evidence of an association of TRAP with childhood asthma incidence involved epidemiologic research where the exposure assessment of TRAP included both traffic-specific pollutants and traffic density measures. It is also possible that the overall lack of agreement relates to the confounding effect of genetics and gene-environment interactions. In fact, a 2009 review of the impact of TRAP on respiratory health called for genetic association studies to clarify how polymorphisms may increase susceptibility to traffic exhaust (Bråbäck and Forsberg 2009). However, for such studies to detect an interaction, a rigorous and accurate exposure assessment of TRAP is crucial. While qualitative critical reviews have suggested that the weight of evidence may indicate possible causal associations between childhood asthma and TRAP (Bråbäck and Forsberg 2009; Health Effects Institute 2010), meta-analyses have demonstrated diverging results. Anderson et al., recently conducted a meta-analyses of the epidemiological evidence published up until 2010 and reported an adverse effect of outdoor air pollution on asthma incidence but no association with lifetime period prevalence of asthma (Anderson et al. 2013a, 2013b). Another meta-analysis showed adverse effects of traffic air pollution on pediatric asthma (Gasana et al. 2012). Although the results of these meta-analyses appear contradictory, a possible explanation could be that the increased risk in incidence reflects the unveiling of the exacerbation of subclinical phenotypes of asthma (Gowers et al. 2012) which highlights the heterogeneity of the disease as well as the need to examine asthma incidence over consecutive time spans. Given that the early life period seems to be a critical time window in the pathogenesis of asthma and allergic diseases, there is a gap for studies following children over time from early life, to preschool and later on into childhood. Epidemiologic results were also inconsistent in studies examining asthma onset in school age participants. For example, positive associations in relation to TRAP exposure have been reported in 20  the Children’s Health Study in California of incident asthma in school age children (Gauderman et al. 2005; Jerrett et al. 2008; McConnell et al. 2010), but no evidence of an association between air pollution and lifetime asthma prevalence in children followed up until age 10 was reported in meta-analysis of birth cohorts from the European Study of Cohorts for Air Pollution Effects initiative (Mölter et al. 2014b). Concomitantly, the discrepancies in results between within-community studies that tend to show positive associations2 (Gruzieva et al. 2013), and between-communities studies that showed null effects (Anderson et al. 2013b; Mölter et al. 2014b) could also be related to a finer-scale exposure assessment that captures individuals living in close proximity to heavy traffic (MacIntyre et al. 2014). These gaps need to be informed.  Critical review of exposure assessment methods for TRAP In determining the role of TRAP in the incidence of asthma and allergic disorders, various exposure assessment approaches have been employed to estimate individual exposures. They vary in terms of complexity, spatial, and temporal resolution. The simplest approach to assign TRAP exposure to individuals in a study has been either the use of a single piece of information to either derive indicator variables (i.e. proximity measures) or to interpolate information. The former approach uses proximity to major roads as an exposure surrogate for the assessment of temporal and spatial distributions of ambient pollution related to motor vehicles, which has been used extensively in air pollution studies (e.g. distance to major roads (Brunekreef et al. 1997)). Similar to the idea of using proximity to roadways as a surrogate for TRAP exposure, traffic intensity has also been used for assigning traffic pollutants in population-based studies (Gauderman et al. 2005; Salam et al. 2004). Proximity and traffic intensity can provide reasonable approximation because (a) pollutants decay studies have shown that NO2, benzene and PAHs                                                      2 It should be noted that meta-analyses showing positive association(Anderson et al. 2013a; Gasana et al. 2012) also focused on within-community exposures 21  are mostly captured between 300m and 500m from roadways (Beckerman et al. 2008), and (b) people tend to live closer to roadways as shown in an  investigation of two large North American cities (Los Angeles and Toronto) (Health Effects Institute 2010) and also more recently in a report showing that 10 million Canadians (32% of the population) live within 100m of a major road or 500 m of a highway (Brauer et al. 2012). An important advantage of proximity measures is the ease of interpretation when used in epidemiologic analyses, which helps with knowledge translation and research dissemination for policy-makers. However, traffic intensity and proximity measures are still crude estimates of exposure as they do not account for wind patterns or land topography and do not incorporate measured values of pollutants.  Another approach is to assign exposure by using surrogate measurements such as NO2 from regulatory air quality monitoring networks. Such monitoring networks have the advantage of capturing temporal trends in pollution concentrations, but do not have good spatial density to capture the smaller scale variations (i.e. within city). Moreover, this approach may introduce exposure misclassification in epidemiologic analyses for pollutants that are spatially heterogeneous, which is the case for TRAP. Geographic Information System-based (GIS) spatial interpolation approaches and statistical techniques have been used to obtain a more precise spatial interpolation of ambient monitoring data (e.g. Kriging approaches) providing further refinement of the proximity metrics. However, these approaches require dense monitoring networks and good input data. Even so, they may still not adequately capture the smaller-scale spatial variations such as those occurring near roadways. Since governmental monitoring air quality stations routinely collect NO2, a large portion of the studies using spatial interpolation have assessed TRAP using NO2 as a preferred marker for TRAP. Other advantages of this marker include its inexpensive measurement by passive samplers, and its correlation with other exposure surrogates such as distance to major road or other pollutant markers such as “soot” (also referred to as black carbon). In the Netherlands, the mean outdoor concentration of NO2 measured at twelve schools was positively 22  correlated with total traffic density on the nearest highway (r = 0.68, p<0.05), and negatively correlated with the natural logarithm of the distance from highway (r = -0.83, p<0.01). Black smoke was positively correlated with NO2 (r = 0.74, p<0.01), and with truck traffic (r=0.67, p<0.05) but not with cars, total traffic, or with the natural logarithm of the distance from highway (Roorda-Knape et al. 1999).  A limitation of relying on pollutant markers from monitoring networks already in place is that these networks are set up for regulatory purposes. They are not designed to capture hotspots or possible within-city variation for which TRAP is an important source, but rather to reflect ambient levels. Since evidence showed that small-scale variability is often as great or even greater than intercity variation for many air pollutants (Briggs 2005; Miller et al. 2007), exposure assessment techniques with poor spatial resolution increase the potential for misclassification because they provide background levels rather than true variability in a given area. Characterizing small area variations in air pollutant concentrations is of utmost importance in epidemiological studies, especially for large studies conducted within a single urban area. This need prompted the use of models capturing spatial variations in air pollution studies. Two general approaches can be distinguished for the air pollution models:  (1) dispersion models, which attempt to simulate the physical (and to some extent chemical) processes involved in transport; and (2) statistical models, which largely ignore the intervening processes, but represent, by empirically defined equations, the relationship between the source and concentrations at the receptor. The latter class of models includes GIS-based models and represents a technique with good spatial resolution that has been used widely in Europe and North America. GIS-based models are also termed land use regression (LUR) because they are based on empirically-derived regression equations linking land use to measured air pollutant concentrations at a set of monitoring sites. In contrast, dispersion models have been used for many years in air quality management but only recently in applied exposure assessment to provide improved spatial exposure estimates (Marshall et al. 2008a). The main limitation of using dispersion models is that all the relevant input parameters need to be included in the model without necessarily 23  gaining a significant benefit in terms of spatial resolution. In a comparison of exposure assessment approaches, urban-scale variations were captured by interpolated background air pollution data collected at monitoring stations and a chemical transport model; whereas neighborhood-scale variations were captured by LUR models (Marshall et al. 2008a). In the same vein, European evaluation studies have indicated that LUR models generally perform as well as or better than dispersion models for NO2 estimates (Briggs et al. 2000; Hoek et al. 2008). Another study reported that LUR outperformed proximity models, kriging, and dispersion models for predicting PM10 (Gulliver et al. 2011). Hence, for environmental epidemiologic studies where effect size is typically small, the ability of LUR models to capture high spatial variability of urban air pollution make it a preferred methodological approach for TRAP exposure assessment.  LUR modeling generally requires a minimum of 40 to 80 monitoring sites, depending on the size and complexity of the study area (Hoek et al. 2008). Other predictors are usually selected to represent traffic-related emissions (e.g. road length, traffic flows) in the surrounding area, and the effect of local topography (e.g. elevation). The predictor variables included in the model vary widely between studies and are usually obtained via geographical information systems (Hoek et al. 2008). In North American studies, LUR has typically achieved predictions for NO2 in the range of R2 = 0.56 to 0.79 (Health Effects Institute 2010). Geographic data collected to build LUR models vary by site, as there is no consensus as to which predictor variables must be included. However, numerous studies have demonstrated an association between GIS-derived information related to exposure and respiratory health effects where there was at least one variable related to road use (either traffic, type or distance) (Hoek et al. 2008; Patel and Miller 2009; Venn et al. 2001). Study areas where there was variation in elevation tended to include a variable related to elevation (Ryan and LeMasters 2007). The variable selection depends therefore on the 24  availability of the data and on the potential sources of pollutant variability in the study area (e.g. shipping emissions, waste burning etc.) A major limitation of LUR models resides in that they have been mostly used to predict NO2 for logistics reasons. As mentioned earlier, passive samplers for NO2 (e.g. Ogawa sampler) are inexpensive compared with other methods to collect traffic pollutants tracers. Alternate approaches such as mobile monitoring that may be applied to additional pollutants have also been utilized. For example, Larson and colleagues conducted mobile monitoring to collect data for a LUR model predicting black carbon. In general, LUR models suffer from a lack of specificity for TRAP.  Given that the relevant timing of exposure for the development of childhood asthma and allergic diseases is yet to be identified, the ability to assign exposures for precise time windows needs to be improved. This is especially critical as LUR models tend to lack temporal resolution. LUR models are developed based on a large number of measurement locations collected in different seasons usually within a year to predict exposures for studies that examine the effect of chronic TRAP exposures (> 1year). These models have traditionally assumed that there is no change in the relative ranking of the exposure of study participants over time. This assumption is only valid if concentrations of pollutants change systematically in all sections of the study area, which is not always the case, leading in turn to misclassification of the study population, especially in the context of large cohort studies spanning long time periods. Few studies have examined how to improve or combine LUR to model temporal variation in air pollution. Su and colleagues proposed a method where LUR included a source area component to predict hourly NO2 concentrations (Su et al. 2008). They showed that this hybrid model is less data demanding than dispersion models, as it requires only meteorological data on wind speed, wind direction and cloud/insolation. Mölter et al. compared three methods to forecast LUR exposure: (1) apply a temporal trend derived from a local background monitor to the concentrations predicted by the LUR model; (2) change the values of the predictor variables to reflect different time periods; and (3) recalibrate existing 25  LUR model, which implies that the regression analysis is re-run with updated values of the dependent variable to reflect different time periods (Mölter et al. 2010). Their results conclude that LUR can provide a relatively good temporal resolution for long-term studies provided that recalibration and adjustment are based on current measured data. More recently, the temporal stability (forward and backward) for the LUR model for the greater Vancouver geographic region was examined and found to provide reliable estimates over a period of seven years measurements (Wang et al. 2013). It was shown that when extended in time, the 2003 LUR model explained 60% variation in 2010 measurements, more than it did in 2003 measurements (59%). Since concentrations and their variability decreased overtime, this allows for a reasonable predictive power in forecasting because it consisted of interpolating within the range of measured exposures whereas backcasting involved extrapolating outside the range of measured values. Limitations associated with LUR estimates of TRAP include: (1) the fact that no study participant stays at a fixed position throughout the study period – although this is true for all methods except measurement –, and (2) the issue of transferability of models over large geographical areas. In large long-term cohort studies, the possibility to transport a model based on one area to another study area needs to be evaluated given the changes in land use characteristics, population densities, and traffic patterns over time. According to the review by Hoek et al. (Hoek et al. 2008) LUR methods can benefit from the inclusion of both a spatial and temporal component. Recent studies have investigated the transferability of LUR models for use in wider geographical areas. However, both Poplawski et al. (Poplawski et al. 2009) and Allen et al. (Allen et al. 2011) showed that even for cities with similar characteristics (Seattle, Vancouver, Victoria and Winnipeg, Edmonton, respectively) it was better to create individual local area LUR models than to have the same LUR model for all areas. At the study area level, people are mobile in the course of the day whereas LUR models predict exposures at one location (usually the home of participants), resulting in misclassification for more mobile study subjects. Time-activity patterns can affect personal exposure. For instance, exposures to pollutants in the transport microenvironment are often highly 26  elevated compared to other microenvironments (Kaur and Nieuwenhuijsen 2009). Data from US travel surveys indicate that in 2001 commuters/adults and children spent an average of 81 and 48 minutes per day in vehicles respectively (Hu 2005), where their exposure to TRAP is likely high. Relative to ambient concentrations measured at monitoring sites, vehicle commuters may receive as much as half the daily exposure in this microenvironment (Gulliver and Briggs 2005; Health Effects Institute 2010). There is a need to adjust LUR model by incorporating mobility information, otherwise the predicted exposure will be inaccurate and may even result in misclassification (Setton et al. 2010). An understanding of factors that influence personal exposures is necessary for a better understanding of the health risks posed by air pollutants in different microenvironments (Kaur and Nieuwenhuijsen 2009). Various approaches have been explored to incorporate mobility into exposure assessments. Through the use of personal sampling, Nethery et al. demonstrated that exposure estimates derived from LUR models were improved when mobility information was taken into consideration (Nethery et al. 2008b). In the case of birth cohort studies, such an approach would be expensive and infeasible. However, questionnaires on time activity patterns can shed light on the amount of time infants spend commuting and in daycare/preschool. Regression models can be improved by incorporating adjustment factors for the portion not predicted by LUR models.  Most of the reviewed literature on assessing exposure to TRAP is concerned with modeling or measuring outdoor pollutants that readily penetrate indoors. Quantifying the PM infiltration efficiency (Finf) in residences can help characterize indoor concentrations and reduce exposure misclassification as there are differences in infiltration between homes and over time (Allen et al. 2006). Misclassification of exposure can affect associations either away or closer to the null because of the different sources of variability associated with infiltration: (1) Studies have shown that Finf can vary from 2 to 10-fold between houses that have the same ambient exposure levels (Allen et al. 2006; Clark et al. 2010a; Hystad et al. 2009) and; (2) Finf can also vary within homes due to seasonal changes. For example, on the Canadian west coast, 27  infiltration varied from 0.49 to 0.72 (Hystad et al. 2009), whereas infiltration in homes in North Carolina varied from 0.50 to 0.63 in heating versus non-heating seasons (Wallace and Williams 2005). In an effort to circumvent the need to assess infiltration, which is deemed impractical in large studies, Hystad and colleagues demonstrated that property assessment data can be useful to model and estimate infiltration based on the value of the structure of a house, also known as improved value. They showed that homes with structure value below the 50th percentile had 15% higher infiltration than homes above the median. Unfortunately, the level of detail on property assessments varies geographically (often by municipality). Clark et al. showed in another property assessment database, for example, that you could not separate the home value into structure and land value (Clark et al. 2010a; Hystad et al. 2009) even within a relatively defined study area.  Finally, TRAP is not solely composed of PM, as outlined previously. This lack of a specific TRAP tracer poses a major challenge for epidemiologic studies of respiratory health effects of air pollution emitted from vehicles. 1.1.3.3 Chemical tracers of traffic-related air pollution in house dust  Despite the variations that occur in dust sampling, house dust measurements have formed the backbone of epidemiological studies of multiple indoor inhalant exposures and asthma. Indeed house dust presents the advantage of providing one matrix for the evaluation of multiple agents related to childhood asthma (e.g. allergens, endotoxins) which is a reasonable proxy for time-integrated exposure (Lioy et al. 2002). While the accumulation of house dust depends on several factors (e.g. infiltration efficiency, pollutant source, cleaning practices, sampling surface), dust concentrations and loadings of pollutants show less variation over time than do indoor air concentrations. Therefore, dust sampling is a particularly useful tool in studies of chronic exposures (Egeghy et al. 2004). The quantification of a TRAP tracer in house dust, as a marker for indoors inhalable hazards and infiltrated pollutants from outdoor origins (e.g. PAHs from vehicles exhaust), would represent a useful and integrated exposure assessment tool. Particulate matter 28  infiltration is the key determinant of indoor concentrations of ambient particles, yet PM is not a specific measure of traffic pollution exposure. A good tracer of TRAP would be a chemical for which the major source is vehicle emissions and that can be measured accurately and at low levels for reasonable cost. One such group of tracers may be the hopanes, a class of organic compounds with 27 to 35 carbon atoms in naphthenic structure (Simoneit 1985). Hopanes are are present in engine oil lubricants, but are not found in gasoline and diesel fuel because they are in the higher boiling fraction of petroleum (Pakbin et al. 2009). Hopanes are thus tracers of primary vehicular aerosols (Delfino et al. 2010), particularly on account of their relative stability and non-volatile nature in the atmosphere (Turlington et al. 2010). Hopanes and steranes can be used to distinguish diesel and gasoline engine emissions from other combustion sources and track personal exposure (Sheesley et al. 2009). These stable species can serve as unique tracers in determining the contribution of diesel and gasoline vehicles to the total concentrations measured in ambient air (Rogge et al. 1993), when sources of fuel (e.g. gas station) and coal combustion are not in the vicinity. In a recent review of organic markers in fine PM,  Lin et al. showed that hopanes are commonly used in receptor model source apportionment studies with the three most abundant hopanes being 17α(H),21β(H),29-norhopane, 17α(H),21β(H)-hopane, and 22,29,30-trisnorneohopane (Lin et al. 2010). Still, the potential for indoor dust as a tool for assessing exposure to outdoor-generated pollution has not been extensively evaluated. According to Turlington and colleagues, there are concerns regarding quality control for hopanes (Turlington et al. 2010). These authors showed that compared to PAHs that have been extensively studied in a variety of matrices (e.g. wipes, vacuumed dust, and urine) with numerous validated methods, there are few choices for hopanes analytical standards. Future research aimed at identifying traffic-pollution constituents likely to have few indoor sources, with the capacity to discriminate near-source traffic from background traffic levels, is needed for epidemiologic studies attempting to link traffic to adverse health outcomes (HEI, 2010). In addition, the possibility to collect 29  such a tracer in house dust offers the possibility to have a time-integrated measure of exposure while capturing the proportion of outdoor TRAP that infiltrates indoors.   1.1.4 Greenness, asthma and allergies The field of research around nature and health has experienced considerable expansion over the past few decades in light of growing worldwide urbanization and related lifestyle changes (Hartig et al. 2014). These factors affect not only air pollution, but also the possibilities of human contact with green spaces, predominantly viewed as beneficial for human health and well-being (Hartig et al. 2014; Lee and Maheswaran 2011). These benefits may act via different pathways, including physical activity (Roemmich et al. 2006), social cohesion, stress reduction and air quality improvement (Figure 1.1). On the other hand, despite the benefits of green spaces, they also pose risks related to human allergic reactions to the airborne plant pollen released during pollination. This coincides with findings which indicate that people living in urban areas are 20% more likely to suffer airborne pollen allergies than people living in rural areas (D’Amato et al. 2007). This situation has been explained by the uniformity of green spaces and the interaction of pollen with air pollutants that potentiate the sensitization response (Cariñanos and Casares-Porcel 2011). In Canada, higher rates of asthma-related hospitalization were shown in relation to an increased load of aero-allergens that can be emitted by certain types of trees suited for urbanized areas (Dales et al. 2004). The risk posed by pollen allergy is amplified in the context of climate change. Climate change has important effects on ecosystems, notably on the potential load of aeroallergens present in the environment. There is now considerable evidence showing that climate change alters the timing, distribution, quality and quantity of allergenic plants and their associated aeroallergens (Beggs 2004; D’Amato et al. 2002, 2010; D’Amato and Cecchi 2008; Shea et al. 2008) which consequently increases the risk of allergies and asthma severity.  30     Figure 1.1: Pathways linking natural spaces to human health (adapted from (Hartig et al. 2014) ). Putative causal pathways in blue, effect modifiers in brown (personal characteristics (e.g. age, gender) and other intrinsic effect modifiers are not depicted for clarity).  With respect to pediatric asthma and allergies, the epidemiologic evidence on the effect of surrounding greenness remains conflicting. Recent studies have provided support to the negative effect of green space exposures via increased allergic load. For instance, in a birth cohort study conducted in New York, asthma at age 7 was significantly increased in those residing in streets with higher tree canopy coverage (Lovasi et al. 2013). Opposite to these results, a recent review on the role of natural spaces (e.g., parks) in children’s health concluded that reduced access to open spaces is related to higher relative prevalence of asthma and allergic conditions in children (McCurdy et al. 2010). This review of evidence is consistent with studies in which exposure to green spaces reduced the risk of asthma (Sherriff et al. 2009) via increased physical activity (Roemmich et al. 2006) or via reduced air pollution exposure (Dadvand et al. 2014; Lovasi et al. 2008). The biodiversity mechanisms (Haahtela et al. 2013) and the revised version of the hygiene hypothesis (Rook 2013) might explain the beneficial role for green spaces with respect to 31  childhood asthma development whereby the reduction in the human-nature interface hinders immunoregulatory mechanisms. Overall, longitudinal follow up studies examining the effect of green spaces on childhood asthma and allergic diseases development are scarce and needed to verify these proposed mechanisms.  1.1.5 Summary of major gaps identified In view of the above critical review of the literature, most studies examining the role of TRAP on childhood respiratory health have used measurements from a few central ambient monitors and/or some measure of traffic as indicators of exposure, disregarding spatial variability and/or factors influencing personal exposure-ambient concentration. There is a need to investigate novel measurement approaches that would be specific to TRAP and/or modeling methods that account for spatial and temporal variation as well as outdoor/indoor infiltration and time-location patterns. On the other hand, asthma and allergic disease development in relation to TRAP research need prospective longitudinal follow-up where phenotypes of asthma and allergic diseases are rigorously characterized and environmental exposures are estimated as early as conception, throughout pregnancy into later childhood. Such follow-up can help clarify critical period of disease incidence both for asthma and for atopy, and define whether various trajectories in the disease development occur following exposure to environmental exposures that co-vary and may interact such as air pollution and green spaces. The latter exposure has been scarcely examined in longitudinal birth cohort. 1.2 Data sources The various investigations reported in this dissertation leveraged two existing Canadian birth cohort studies. A description of the study populations, outcome measures and exposure assessment methods are described below. 32  1.2.1 Canadian healthy infant longitudinal development (CHILD) study 1.2.1.1 Study population and outcome data The Canadian Health Infant Longitudinal Development (CHILD) study is a large (n=3500 mother/infant pairs) prospective birth cohort that was designed to determine the role of environmental factors and their interactions with genetic and host factors in the development of allergy and asthma in children. Pregnant mothers from four Canadian cities (Vancouver, Edmonton, Winnipeg, and Toronto) were recruited between 2008 and 2012. Surveys of environmental exposures using questionnaires and home assessments were complemented by extensive environmental and biologic sampling (Moraes et al. 2015; Takaro et al. 2015).  Inclusion and exclusion criteria for mothers and infants are shown in Table 1.1. Efforts were made to recruit from a representative general population to allow identification of environmental risks and novel genes by avoiding the selection of an at-risk population. Infants born prematurely or with respiratory distress syndrome were excluded as these conditions can independently increase the development of wheeze, asthma and allergic respiratory disorders. Similarly, families who turned to in vitro fertilization were excluded since potential environmental effects and epigenetic changes in the embryo developing in vitro could not be assessed. Finally, children who spent less than 80% of their time in the principal residential address were excluded because of the inability to model their exposures (these do not include children spending the equivalent of a work day in daycares). Participation of the father was encouraged but not mandatory for recruitment. All participants provided informed consent at the time of enrolment. Table 1.1: Inclusion and exclusion criteria for participants’ recruitment in the CHILD study Inclusion Criteria Exclusion Criteria Pregnant women aged >18 years (>19 years in Vancouver). Children born with major congenital abnormalities or respiratory distress syndrome (RDS). 33  Inclusion Criteria Exclusion Criteria Residential proximity (<50 Km) to participating delivery hospital. Expectation of moving away from a recruitment center within 1 year of recruitment. Ability to read, write and speak English. Children of multiple births. Willing to donate cord blood. Children resulting from in vitro fertilization. Planning to deliver at a designated recruitment center participating hospital. Children who will not spend at least 80% of nights in the index home. Infants born at or after 35 weeks.   To enable the examination of potential cause-effect relationships relevant to the development of pediatric allergy and asthma, the CHILD study is defined by detailed phenotyping, based on clinical data and objective measurements of atopy and lung function, together with assessment of a diversity of pre- and post-natal environmental exposures and biobanking of multiple biological and environmental samples including blood, breast milk, urine, stool, nasal swabs and house dust for subsequent assays of chemical and biological agents. All CHILD participants followed the same visit schedule (Table 1.2), except for some tests and biological samples that were city-specific. In particular, families recruited in Toronto had an extensive set of early novel lung development characterization examinations (Pulmonary Function Tests at 3, 12 and 18 months of age). Similarly, only Vancouver mothers were administered a particular psychological scale (the Life Stress Interview) at 36 weeks of pregnancy and when the child was one year old. Finally, only Winnipeg children provided venous blood at the three year clinic visit.    34  Table 1.2: CHILD study schedule from recruitment to age 5 Visit 18 Week 36 Week BIRTH 3 Month home  6 Month 1 Year Clinic Visit  1.5 Years 2 Years 2.5 Years 3 Years Clinic  4 Years 5 Years Clinic  Questionnaires  Mother Profile/Residential history √     √   √   √   √ √ √ Mother Health  √                     √ Mother Nutrition  √                     √ Mother Medications  √                       Mother Psychosocial √ √                   √ Parenting stress           √   √   √ √ √ Father Health  √                       Socio-economic Status (SES) √                       Child Delivery Chart      √                   Child Health        √                 Child Nutrition        √                 Child Medications        √                 Child Clinical Assessment                          Home Environment Survey √     √ √ √ √ √ √ √ √ √ Home Assessment done by RA       √                 Tests Mother Skin Prick Test           √             Mother Spirometry            √             Father Skin Prick Test  √                       Father Spirometry  √                       Child Skin Prick Test            √       √   √ Child eNO                       √ Child Pulmonary Function Test                        √ Samples Cord Blood     √                   Mother Breast Milk        √                 Mother Venous Blood  √         √             Parent Buccal Swab           √             Father Venous Blood  √                       Chid Venous Blood            √             Child Nasal Swab       √   √           √ Chidl Buccal Swab       √   √             Child Urine        √   √           √ Child Meconium/Stool        √   √             Home Dust Collection        √                  35   The analysis in this dissertation took advantage of the availability of skin prick tests administered to the mother/child dyad one year after birth to examine early infancy risk of sensitization with regard to air pollution. During the clinical assessment where the skin allergy test was performed, trained pediatric nurse or research assistant visited each participant. The skin allergy testing (Appendix A), performed on the arm or the back, used standardized ALK (Horsholm, Denmark) inhalant aeroallergens (tree, grass and weed pollens, cat, dog, Dermatophagoides pteronyssinus, D. farinae, cockroach, Alternaria, Cladosporium, Aspergillus, and Penicillium) and common food allergens (cow’s milk, egg white, soybean and peanut). Individual weal sizes for allergens or positive (histamine 0.1%) and negative (glycerin) controls were determined by the mean of the longest diameter and its perpendicular. After adjusting for any response to the negative control, atopy is characterized dichotomously, defined by 1 or more positive skin tests with mean weal diameter ≥2 mm above the negative control response.  1.2.1.2 Environmental factors and covariates  The methodology describing environmental assessments (Table 2) was reported in detail elsewhere (Takaro et al. 2015). In brief, indoor environmental exposures were repeatedly assessed by house and content questionnaires from pregnancy to age 5 years including details of indoor behaviors. For the purpose of this work, tools were administered during pregnancy up to 1-year post partum. When the child was around 3 months of age, a detailed walk-through assessment (Appendix B) with dust sampling for multiple allergens and chemical pollutants, including hopanes (Chapter 2), was conducted, along with global positioning system (GPS) tracking that was aimed to complement the self-reported residential history addresses used to identify the index home of each participant. Time activity patterns of the child and parents were tracked by means of questionnaires and included time spent in transit, and mode of 36  transport. Second-hand tobacco smoke exposure was assessed by repeated questionnaires after enrollment (i.e. during pregnancy, after birth (around 3 months), and when the child was one year of age).  Along with the assessment of environmental risk factors that were delineated in Section 1.1.2, other working groups within CHILD developed questionnaires to assess known asthma risk factors, namely diet, psycho-social environment characteristics, and birth characteristics. Table 1.3: Principal exposure assessment methods used in the CHILD Study to characterize the physical environment during pregnancy and infancy periods                   Method  Exposure example Questionnaires Physical home assessment Home dust samples GIS model  Common allergens    (includes pets and pests) √ √ √3   Secondhand smoke √ √    Dampness √ √    Mold  √ √ √3  Traffic Related Air Pollution √  √ √  1.2.1.3 Traffic air pollutants LUR models of nitrogen dioxide have been developed for several Canadian cities, including Windsor (Wheeler et al. 2008), Montreal (Gilbert et al. 2005), Edmonton and Winnipeg (Allen et al. 2011), Vancouver where models were developed for different TRAP tracers  (Henderson et al. 2007; Larson et al. 2009), and Toronto (Jerrett et al. 2007). CHILD cities have a specific LUR model for NO2 (Table 1.4) which is commonly used as a TRAP marker (Chapter 3) and for which model development methodology was virtually identical for Vancouver, Winnipeg and Edmonton.                                                       3 Allergens and Mold in house-settled dust were not examined as part of this dissertation. They are merely indicated here to demonstrate the breath of physical exposure assessment methods in CHILD. 37  Table 1.4: LUR models for nitrogen dioxide (variables4 , estimated annual means for all participants and standard deviation (sd)). City  Land Use Regression Model  R2 Mean 5 (sd) (µg/m3) LUR error (sd)6 Edmonton(Allen et al. 2011) NO2 = 16.60 + 0.02(IND_1500) + 0.30(RD1_1000) + 12.01(RD2_50) + 4.66(RD.100) + 0.06(WTR.1000) + 0.58(Distance City Center) 0.81 22.2 (9.1) 0 (3.0) Winnipeg(Allen et al. 2011) NO2= 4.11 +  0.007(IND.2000) +  0.33(IND_200)+  7.14(HM_50) + 5.64(RD.75)+ 0.09(POP_2500) + 0.18(Y) 0.77 8.6 (5.0) 0 (2.6) Vancouver(Henderson et al. 2007) NO2= 42.6 + 10.5(RD1_100) + 0.274(RD1_1000) + 4.24(RD2_200) + 0.074(POP_2500) + 0.116(COM) – 0.02(ELEV) – 0.591(X) 0.76 21.2 (8.9) 0 (5.2) Toronto(Jerrett et al. 2007) Log(NO2) = 8 + 0.18(RD1_200) + 0.6(RD2_50) + 0.0016(IND.750) + 8.3*10-5 (DC2000) - 8*10-6(X) + 0.13(D_WIND.1500) + 0.001(TRAF.500) 0.67 27.1 (8.9) 1.9 (4) 0 (0.1) 1.4 (3.1)  The second metric that was examined is hopanes in settled house dust. Details on the methodology and protocol adopted for the collection and analysis of hopanes are found in Chapter 2, which is dedicated to the examination of the utility of this class of chemical as a TRAP tracer. Standard operating procedures are annexed in the appendix B.                                                      4 All the buffer zones radii for the derived land use variables are indicated in suffix with the radius length in parenthesis. IND: industrial; WTR: Water; RES: Residential; COM: commercial; RD1: Highway; RD2: Major Road; RD: all roads; HM: Highway or Major Road; ELEV: Elevation; POP: population density; DC: density of dwellings; TRAF: Traffic counts; D_WIND: downwind of major expressways; Y: Latitude; X : longitude. 5 The temporally adjusted annual concentrations and standard errors in µg/m3are presented for each city.  6 For all cities except Toronto, we show the leave-out one error estimate with the standard deviation around the estimate. For Toronto, we show the results (absolute and relative difference) of three cross-validations using 70% randomly selected sites to predict the remaining 30% 38   1.2.2 Border air quality study (BAQS) Part of the research described in this thesis leveraged the existing Border Air Quality Study (BAQS). BAQS includes a series of research projects aimed at evaluating the health impacts of air pollution in the Georgia Basin / Puget Sound Airshed, located in Southwestern British Columbia (BC), Canada. 1.2.2.1 Study population and area The work presented here focuses on the study area encompassing the municipalities of the Greater Vancouver Regional District (GVRD) with the addition of the neighbouring municipalities of Abbotsford, Chilliwack, and Mission, which are located in the Fraser Valley (Figure 1.2). The addition of the neighboring municipalities to the GVRD study area is subsequently referred to as GVRD plus. Recent predictions for the GVRD municipalities suggest a population of 4 million inhabitants by 2040 (Greater Vancouver Regional District), that is likely to result in urban sprawl, increased transportation and energy consumption. This will likely impact the region’s air quality since the industrial and transportation emissions (motor vehicles and marine transport) are important sources of air pollution year-round.  The birth cohort consisting of all children born in the study area during the 1999 to 2002 period comprised a total of 68,326 children who were successfully linked to administrative health databases and geospatial exposures beginning at conception. The study population is described in more details below (sections: 1.2.2.2 to 1.2.2.4).  39   Figure 1.2: Study area of GVRD plus within the Border Air Quality Study 40   1.2.2.2 Defining asthma using administrative databases Administered by Population Data BC (https://www.popdata.bc.ca), various population-based and health records databases in British Columbia are linked, including Medical Services Plan records (British Columbia Ministry of Health 2009b), hospital separations records (British Columbia Ministry of Health 2009a), births and deaths records through Vital Statistics (British Columbia (BC) Vital Statistics Agency 2009a, 2009b), as well as the BC Perinatal Data Registry (Perinatal Services BC 2009) and Census data (Statistics Canada 2001).  The Medical Services Plan records include information on every outpatient physician visit and medical procedure billed to the provincial medical plan. Enrolment in the provincial plan is mandatory for residents of the province (BC Ministry of Health 2007) and therefore it captures data at a population level. Physician billing claims to the medical plan require an ICD (International Classification of Diseases) code indicating the primary reason for the encounter as well as administrative details, including physician, date, location, and patient’s address. Personal Health Numbers that are unique to individuals can be used to link this data to other provincial databases, including the BC Perinatal Database Registry. This database collects detailed information on every birth in the province, including maternal and birth characteristics. These linked data sources were used to identify and follow the study population, create a residential history file, assess asthma diagnosis via physician visits and hospitalizations throughout the follow up period, and assess risk factors. All data were linked based on personal health number, except for census data linked on census dissemination area of residential location.  Administrative data provide consistently high rates of follow-up that are not always achievable with primary data collection. Therefore, this approach for epidemiological research offers not only an efficient and economical means to examine disease incidence and trends over long periods but also a population-41  based record of health information that is of a scale virtually unfeasible in birth cohorts and that circumvents the inherent selection bias in survey-based research reliant on self-reported health outcomes. Several studies in Canada have evaluated the use of administrative records to identify asthma cases for health research. The relative accuracy of asthma coding in this context was demonstrated in multiple Canadian provinces including Quebec (Blais et al. 2006; Labrèche et al. 2008), Manitoba (Lix et al. 2008), and Ontario (To T et al. 2007; To et al. 2006) where administrative data used for case ascertainment were compared to either physicians’ charts or patients self-reports. In Quebec, the investigators verified administrative asthma diagnoses against chart recorded asthma diagnoses in individuals aged 16-80 and demonstrated that the provincial databases were a valid measure of asthma via physician’s billing. In Manitoba, Lix and colleagues showed that the case definition algorithm affects the validity of findings compared with self-reports. For children aged 12-18, specificity remained high for all algorithms (92-97%) but sensitivity varied significantly (from 16-87%) depending on the number of years of data examined. The sensitivity increased with the number of years of data available for analysis.  To et al. examined asthma diagnoses in the Ontario billing databases for children aged 0-18 years. They found that the diagnosis entered by the primary care physician in a patient’s chart was very reliably recorded in the administrative database, with 99% overall agreement (To et al. 2006). The recorded diagnoses were also validated by experts blinded to physician’s recorded diagnosis against chart review. These charts agreed well with the administrative data (overall sensitivity was found to be 91% and specificity 83%). The sensitivity was found to be lower (75%) for young children (aged 0-2), but rose to 96% for the 3-5 year old group. Specificity remained consistent at or above 83%. All together, these studies indicate that administrative databases, especially if available over several years, are a sensitive and specific measure of physician-diagnosed asthma. 42  1.2.2.3 Exposure estimates: air pollution  Air pollution exposure was estimated using proximity to roadways, interpolated measurements of regulatory monitoring data, and LUR modeling, (Brauer et al. 2008; Henderson et al. 2007; Larson et al. 2007).  Exposures were assigned at the level of 6-digit postal codes for residential address. This corresponds to one edge of a block in urban areas, but is larger where population density is lower.  The regulatory monitoring network consists of daily measurements at 24 monitors for O3, 22 for NO and NO2, 14 for SO2, 19 for CO and PM10. Exposures were assigned according to inverse distance weighting (IDW), in which the daily values at the three closest monitors within 50 km were weighted by their inverse distance (1/d) to the postal code of interest.   Annual average air pollution LUR models were also developed for the region using measurement data from targeted intensive sampling campaigns. Modeling was used to develop high-resolution (10 meters) maps of NO, NO2, PM2.5 and black carbon using GIS data on road density, population density, elevation, and type of land use. LUR models have improved spatial resolution as compared with the monitoring network approaches, but lower temporal resolution due to sampling data from a single year. For each LUR model, the corresponding monitoring network data were used to temporally adjust LUR estimates (Times Series Forecasting System, version 9; SAS Institute Inc., Cary, NC, USA). Then, these were adjusted to estimate monthly average concentrations (Brauer et al. 2008; Henderson et al. 2007).  For proximity to roadway measures, residential postal codes were defined as being within 50 m or 150 m from highways and major roads (DMTI ArcView street file dataset for BC, Canmap Streetfiles, v2006.3, 2006).  43   For all exposure metrics, an average exposure was calculated for gestational life up until birth. These exposure estimates were weighted by the time spent at each residence to incorporate individual residential histories derived from BC Ministry of Health data.   1.2.2.4 Exposure estimates: surrounding greenness Satellite-derived Normalized Difference Vegetation Index (NDVI), a biomass density indicator, was used to assess a continuous measure of greenness across the Metropolitan area of Vancouver, BC. NDVI data ranges from -1 (water) to +1 (dense vegetation), with values close to zero indicating barren areas of rock, sand or snow based on land surface reflectance of visible (red) and near infrared parts of spectrum(Weier and Herring 2000). To assign greenness measures to study postal codes, all cloud free images from Landsat Enhanced Thematic Mapper Plus (ETM+) of the Vancouver region were downloaded from 1999-2002. Average greenness values were extracted for 100 m areas around residential postal codes and yearly greenness values calculated. The small 100 m area was used to assign residential greenness exposure as these distances have been used in prior studies (Dadvand et al. 2012a, 2012b), and are at a similar spatial scale to within-city air pollution variation.  1.3 Objectives  Fifty-four percent of the world’s population lives in urban areas, a proportion that is projected to rise to 70% by 2050 (Talukder et al. 2015). Despite technological advances in terms of vehicles emissions and the decline in air pollution levels seen in Canadian cities such as Vancouver (Brauer et al. 2012; Brook et al. 2014), the health burden caused by the transport demands is likely to pose serious public health challenges above and beyond today’s concerns. Given the ubiquitous nature of traffic-related air pollution and the fact that a “safe concentration” has yet to be identified, there is a clear need for better risk characterization in order to put proper control and mitigation strategies in place. Such risk 44  characterization is dependent upon several aspects including the ability to assess personal exposures, both accurately and precisely.  This work was motivated by the need to extend/improve existing methods of exposure assessment to TRAP that would account for modifiable factors (e.g. time-activity patterns, potential indoor sources of TRAP markers) in order to reduce exposure misclassification and strengthen the epidemiological evidence on the role of outdoor and indoor air pollutants in pediatric allergic diseases and asthma development, and in particular the role of TRAP.  The overall goals of this research were twofold: first to improve exposure assessment of air pollution for large birth cohort studies of childhood respiratory disease, and secondly to determine the impact of TRAP the development of asthma and allergies both in early years and later in childhood. A specific objective related to the first goal was to describe the methodology and provide descriptive information on TRAP exposures in the Canadian Health Infant Longitudinal Development study (CHILD).  The proposed dissertation is articulated around the following research objectives that stem from the identified gaps in the literature reviewed above and that help answer this dissertation’s main objectives: 1- To examine associations between hopanes concentration profiles in outdoor air and hopanes in house dust, and between GIS predictors and NO2 ambient exposures at CHILD participating homes where dust samples were collected (Chapter 2). 2- To develop and apply methods for assessing traffic-related air pollution exposure in a multi-center birth cohort study which account for spatial and temporal variation as well as time-activity patterns and home infiltration efficiency (Chapter 3). 3- To evaluate whether traffic-related air pollution exposure is associated with allergic responses to common inhalant and food allergens (Chapter 3) by applying the exposure estimates in Objective 2 above in the analysis of atopy development in infants at age 1. 45  4- To examine associations between traffic-related air pollution and the development of asthma among children followed from birth to ten years of age in a population-based birth cohort study and investigate when these associations (if any) with new onset of asthma occur during preschool (0-5 years old) and school age years ( ≥ 6 years old)  (Chapter 4).  5- To evaluate the effect of surrounding residential greenness on new asthma onset in children and its potential role in modifying the TRAP-asthma incidence relationship (Chapter 4). 6- To examine asthma trajectories over ten years of follow up using a data-driven approach to clarify time of asthma onset and persistence in relation to neighborhood air pollution and greenness exposures (Chapter 5). 1.4 Dissertation structure  The layout of this thesis conforms to the manuscript-based thesis guidelines of the University of British Columbia. Chapters 2 and 3 have been published in a peer reviewed journal, and Chapters 4 and 5 have been submitted for publication.  Chapter 2 is a cross-sectional between-communities investigation of the utility of hopanes as a traffic-related air pollution marker that incorporates the infiltration of ambient pollution inside the homes. Chapter 3 is a longitudinal study in which the potential effects of traffic-related air pollution exposure on allergic health outcomes are explored in the Canadian birth cohort, CHILD. Complementing its previous chapter, chapter 4 relies on an unselected population as it longitudinally examines the influence of traffic-related air pollution in a population-based study (BAQS) where all children are followed up for 10 years. Chapter 5 extends the previous one by identifying group memberships in the same population based on their disease status and recurrence, then examining the influence of traffic-related air pollution on the identified trajectories. Finally, the conclusions and implications of this dissertation, as well as its strengths and limitations, are discussed in Chapter 6 where recommendations for future research are provided.  46   Chapter 2: Assessing hopanes as an exposure assessment tool for TRAP exposure7  2.1 Introduction The burden of air pollution from traffic on morbidity is also well documented with a variety of negative respiratory and cardiovascular effects (Brauer et al. 2002; Gan et al. 2010) as well as adverse pregnancy outcomes (Brauer et al. 2008) and lung cancer (Beelen et al. 2008; Vineis et al. 2006). According to a recent comprehensive review (HEI, 2010), there is sufficient evidence to infer a causal role for TRAP in the exacerbation of asthma in children and suggestive evidence that it may play a role in the onset of asthma in children (Health Effects Institute 2010).  A number of pollutants (e.g. CO, NOX, and PM components) that are routinely measured at fixed regulatory monitoring sites can be used to represent exposure to TRAP. However, regulatory monitoring data cannot capture the fine-scale spatial pollutant gradients associated with vehicle emissions. Most of the recent epidemiological studies assessing TRAP have improved on such earlier exposure assessment approaches by using more elaborate methods with higher spatial resolution to provide individual-level exposure estimates. These methods generally estimate different surrogates of the traffic mixture (e.g. NO2, black carbon) derived from dispersion or land use regression (LUR) models (Hoek et al. 2008). Despite the advances in TRAP exposure assessment, none of the surrogate pollutants measured or modeled are specific to vehicle emissions.                                                       7 A version of this manuscript has been published: Hind Sbihi, Jeffrey R Brook, Ryan W Allen, Jason H Curran, Sharon Dell, Piush Mandhane, James A Scott, Malcolm R Sears, Padmaja Subbarao, Timothy K Takaro, Stuart E Turvey, Amanda J Wheeler and Michael Brauer. 2013. A new exposure metric for traffic-related air pollution? An analysis of determinants of hopanes in settled indoor house dust. Environmental Health 2013, 12:48  doi:10.1186/1476-069X-12-48 47  In addition to the lack of specificity, all these methods characterize ambient levels and do not consider indoor infiltration. Since individuals in North America spend an average of 87 per cent of their time indoors (Klepeis et al. 2001), (Leech et al. 2002) and many outdoor pollutants readily penetrate indoors, a significant proportion of total exposure to outdoor-generated pollutants occurs indoors. For example, quantifying the PM infiltration efficiency (Finf) in residences can help characterize indoor concentrations and reduce exposure misclassification (Allen et al. 2006) since Finf can vary 2 to 10- fold between houses that have the same ambient concentration levels (Allen et al. 2006; Clark et al. 2010a; Hystad et al. 2009). Unfortunately, methods for estimating Finf in residences require indoor and outdoor pollution sampling, which makes estimating Finf in large epidemiological studies virtually impossible. To overcome this challenge, prediction models of Finf have been developed (Clark et al. 2010a; Hystad et al. 2009). While they have shown promise, these models have generally been developed for individual cities using relatively small sample sizes and therefore may not be transferable to other locations. Hence, current approaches to assess individual TRAP exposures (LUR, dispersion model, geostatistical methods) have two consistent limitations:  (1) TRAP surrogates are based on non-specific pollutant measures, (2) modeled estimates are capturing concentrations outside the home while most 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 dust sampling, this type of sampling has formed the backbone of epidemiological studies of multiple biological agents (Arrandale et al. 2010). Indeed, house dust presents the advantage of providing one matrix for the evaluation of multiple agents which is a reasonable proxy for time-integrated exposure (Lioy et al. 2002). While the accumulation of house dust depends on several factors (e.g. infiltration efficiency, pollutant source, cleaning practices, sampling surface), dust concentrations and loadings of pollutants show less variation over time than do indoor air 48  concentrations, therefore, dust sampling is a particularly useful tool in studies of chronic exposures (Egeghy et al. 2004). Using house dust as a marker for indoor inhalable hazards and infiltrated pollutants of outdoor origin (e.g. polycyclic aromatic hydrocarbons (PAHs) from vehicles exhaust) would represent a useful and integrated exposure assessment tool. A good tracer of TRAP in dust would be a chemical: 1) for which the major source is vehicle emissions, 2) for which emissions are correlated with other motor vehicles constituents, 3) that can be measured at low levels for reasonable cost, and 4) that can be measured with accuracy and stability.  One such group of tracers may be the hopanes, a class of organic compounds with 27 to 35 carbon atoms in a naphthenic structure (Simoneit 1985). Hopanes are not found in gasoline and diesel fuel because they are in the higher boiling fraction of petroleum, but are present in engine oil lubricants (Pakbin et al. 2009). Hopanes are tracers of primary vehicular exhaust aerosols (Delfino et al. 2010), particularly on account of their relative stability and non-volatile nature in the atmosphere (Turlington et al. 2010). Schauer et al. showed that hopanes and steranes could be used to distinguish diesel and gasoline engine emissions from other combustion sources (Schauer 2003). These stable species can serve as unique tracers in determining the contribution of diesel and gasoline vehicles to the total particulate matter concentrations measured in outdoor air (Brauer et al. 2002; Rogge et al. 1993). In a recent review of organic markers in fine PM,  Lin et al. showed that hopane monomers  are commonly used in receptor model source apportionment studies with the three most abundant species being 17α(H),21β(H),29-norhopane, 17α(H),21β(H)-hopane, and 22,29,30-trisnorneohopane ( Lin et al. 2010). The potential for assessing exposure to outdoor-generated pollution in indoor dust has not been evaluated. Measurement of hopanes in house dust offers the possibility to have a time-integrated measure of exposure while capturing variability in the infiltration of TRAP, circumventing the need to indirectly account for infiltration. Therefore we evaluated the association between hopane concentrations in house dust samples to outdoor TRAP surrogates and hopane concentration profiles in outdoor air. 49   2.2 Methods Our overall goals were to determine (1) whether the hopane mixture in house dust had similar composition as that in outdoor air and (2) the relationship between hopanes in settled house dust with predictors of TRAP spatial variability. We utilized indoor dust measurements from five Canadian cities (Vancouver, Edmonton, Winnipeg, Toronto and Windsor), in order to ensure sufficient variability in hopane levels. Specifically, we conducted a city-level analysis where both city-specific and harmonized LUR variables across all cities were examined. These analyses also included covariates identified in housing characteristics surveys that were administered in the different studies used for this investigation. 2.2.1 Population  Samples were collected in three separate studies, briefly described here, in which house dust was collected from inside homes of study participants:  • (1) the Canadian Healthy Infant Longitudinal Development (CHILD) study is a prospective longitudinal, population based birth-cohort study that has enrolled 3650 families from Vancouver, Edmonton, Winnipeg, and Toronto between 2009 and 2012. Homes that (i) underwent home assessment when the child was at an age of 3-months; (ii) completed the questionnaires on environmental factors, and (iii) had dust samples with sufficient dust mass for the analysis of several agents (endotoxins, β-glucans, and hopanes) were selected while ensuring balanced sample representation from the four CHILD cities. Thus, 120 homes analyzed for the suite of hopane monomers by December 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 larger survey of 5,619 students who completed a screening survey for respiratory disease (Dell et al. 2010). Within this nested study, a sub-sample of 50 homes were inspected in 2006/2007 and underwent measurement of 50  indoor/outdoor concentrations of traffic related pollutants. From these, only 24 homes, included in this study, with sufficient dust mass for the analysis of allergens (Der p, Der f, Ergosterol and Glucans) were also analyzed for hopanes (Loo et al. 2010);  • (3) during 2005/2006, Health Canada and the University of Windsor conducted a personal exposure study in Windsor (Wheeler et al. 2011) (the Windsor Ontario Exposure Assessment Study, WOEAS), in which in which  48 households  were randomly recruited from the larger Windsor Children’s Respiratory Health Study (Dales et al. 2009), and where preference was given to spatially distributed households across Windsor. From these households, all homes with sufficient house dust mass were selected (n=27) to examine the hopanes levels in house dust.  2.2.2 Hopanes  2.2.2.1 Dust Samples: collection and analysis  House dust samples were collected from the living rooms in all the homes included in the study. Participants were asked not to vacuum during the week prior to the home visit. Sampling was conducted by trained technicians who were instructed to measure the sampling area, note the type of surface and collect a pre-determined amount of dust.  The WOEAS and TCHEQ sampling protocols were similar as technicians vacuumed a 4m2 section of floor for a period of 4 minutes or until at least one gram of dust was collected and used high volume devices. In WOEAS, settled dust was collected using the High Volume Surface Sampling System (HVS3) vacuum (Wheeler et al. 2011), while TCHEQ used the Shop-Vac vacuum (Model: QAM70, 7.0 Amps), another high volume device, equipped with Dust Sampling Socks (X-Cell 100, Midwest Filtration, Cincinnati, OH, USA). In CHILD, house dust samples were collected using a standardized consumer model vacuum cleaner (Sanitaire, model S3686) fitted with a dust collection device designed especially 51  for the CHILD study with the goal of increasing the collection efficiency without having to vacuum the entire area. This modified collector included slots for two nylon filter thimbles, thereby doubling the filtration surface and was constructed from machined aluminum, and outfitted with Teflon wheels to prevent marring of flooring, and to maintain the collection slot at a fixed distance from the test. The sample was taken from a 2 m2 area by making seven passes of the nozzle over adjacent swaths of flooring. Fresh weights of dust samples were taken, after which they were sieved to 105 µm and this fraction reweighed. These sieved fractions were aliquoted and frozen at –80° C pending further analysis. Hence, all dust samples were sieved into <150µm size fractions for analysis. The sieved fractions were aliquoted and frozen at -80°C pending further analysis at the Environment Canada laboratory. Extraction in an isooctane solution was conducted with an ASE 200 (Accelerated Solvent Extractor) followed by solvent reduction using a Zymark TurboVap. Recovery standards were added to the dust/solvent matrix before extraction and blow down. A suite of organic compounds were quantified by tandem Gas Chromatography-Mass Spectrometry, including eleven hopane monomers. The final dust-related metric for each of the individual hopanes and the sum of all 11 monomers was expressed as the concentrations per gram of sieved dust (ng/mg), thereby correcting for differences in the total amount of dust collected in each sample.  2.2.2.2 Outdoor hopane measurements  The composition of hopane mixtures, expressed as the abundance of the highest concentration monomer (17α(H), 21β(H)-Hopane) relative to the sum of the concentrations of all 11 measured monomers, was compared between available PM2.5  outdoor air samples in Vancouver, Edmonton, Toronto and Windsor with house dust samples for the same cities. In all cities, one 24-hr ambient PM2.5 sample was collected at Environment Canada national monitoring network sites (Dabek-Zlotorzynska et al. 2011)  within the 52  same week in the months of January, April, July and  October 2010 . The same suite of hopane monomers available in the dust samples were quantified at the same Environment Canada Laboratory by thermal desorption gas chromatography mass spectrometry (Graham et al. 2010) from punches of archived pre-fired quartz filters. Dust and air samples were matched by city and season. 2.2.3 Geographic predictor variables  Harmonized geographic data were derived to allow for pooled analysis of all dust hopane measurements from all five cities where samples were collected. We generated 30 variables in 5 categories that are often used in development of LUR models for TRAP (Dijkema et al. 2011; Hoek et al. 2008). Subcategories were generated to characterize the street network, land use, and population density within circular buffer sizes where the radius was set to represent close, medium and large geographical areas around each home where the house dust sampling was conducted (Table 2.1). Highways and major roads were defined by standard road classification categories (DMTI Spatial Inc., Markham, Ontario), with categories 1 (expressway), 2 (principal highway), and 3 (secondary highway) all considered highways (RD1), and category 4 as major roads (RD2). We also examined land use, elevation relative to sea level and the distance to the nearest features within the street network.  All variables in each category were derived from a single spatial dataset in vector format. Input files for the Road Length and Land Use, were taken from the 2006 DMTI Spatial (Markham, Ontario) data files. Population Density categories were generated from the 2006 census distributed by Statistics Canada and converted into point files at the block level. Digital Elevation Data was obtained from GeoBase in raster format at the municipal level.  All input files were manipulated in ArcGIS 10.1 (ESRI, Redlands, CA) to produce variable layers in raster format at 10m resolution, except for the digital elevation model where the finest available resolution was 30m. From the latter data, relative elevation was defined as the mean centered city-specific elevation. 53   Table 2.1: Harmonized GIS variables across TCHEQ, WOAES, and CHILD cities   We also extracted city-specific variables that had previously been extracted and used in the development of LUR models for NO2 in each of the cities (Allen et al. 2011; Henderson et al. 2007; Jerrett et al. 2007; Wheeler et al. 2008) (Table 2.2). Since these LUR variables had been used to explain variability in outdoor NO2 in these cities we therefore expected that they would explain variability in dust hopanes concentrations.   Category (Number of variables) Description Sub-Category Buffer radii (m) Source/Type Road Length (8) Total length of two road types RD1 (Highways) RD2 (Major Roads) 50, 100, 500, 1000  DMTI Road Network (Polyline) Land use (12) Total area of different land use types (ha) COMM (commercial) OPEN PARK INDUS (industrial) 100, 500,1000 DMTI spatial data  (polygon) Distance to nearest feature (6) Distance to nearest road type (m)   Dist_RD1 Dist_RD2  DMTI Road Network (Polyline) Distance to nearest land use type (m) Dist_Comm Dist_Open Dist_Park Dist_Indus  DMTI spatial data  (polygon) Population density (3) Density of the population (persons/hectare) POPDENS 100, 1000, 2500 Block level census data (point file) Geographic position Elevation (m) ELEV  Geobase DEM (raster) 54   Table 2.2: City-specific GIS variables and buffer sizes extracted from LUR model surfaces. City/ Cohort Road metrics (m) Traffic density Other Edmonton CHILD Length of all roads  (100,250, 500, 1000m)  Distance to city center (m) Water area Winnipeg  CHILD   Y coordinate centered at city center. Vancouver CHILD  Length of Truck roads (100,250, 750, 1000m) Automobile  density  Truck density (100, 250, 750, 1000m)  Toronto TCHEQ Distance to nearest expressway   Length of expressways  (100,500, 1000, 2000m)   Windsor WOEAS Distance to nearest  local road Distance to  nearest class 1, 2 collector  Distance to nearest class 1, 2  arterial  Distance to nearest  highway Distance to nearest scenic drive  Distance to Ambassador bridge (m) Distance to Windsor/Detroit Tunnel (m)   2.2.4 Questionnaires  We also included data from questionnaires delivered in each of the indoor measurement studies on housing characteristics and lifestyle factors, which may be related to indoor hopane variability and/or infiltration (Table 2.3).     55   Table 2.3: Descriptive summary of questions found (as shown with a check mark) in the questionnaires delivered during home visits, recoded for analysis in the pooled investigation of hopanes in dust and land use determinants of traffic pollution. Question type CHILD TCHEQ WOEAS Emissions sources within 100m of the home Factory  Gas Station Parking  Construction Site √  2% 11.3% 15.6% 23.5% × × Shoe removal  Yes No √ 94% 6% × × Type of Floor  mixed  smooth  carpets √ 4% 21% 75% √  83% 17% √ 4% 36% 60% Cleaning frequency  Rarely Moderately Frequently √ 14% 80% 6% √ 4% 71% 25% √  56% 44% Window usage/type Usually open/sheer Covered with blinds/curtains Sealed Opened daytime/ closed night Other √ 15% 42.5% 34%  2% √ 37.5% 14.8%  14.8% 14.8% √ 33.3% 63%  3.7% 0% Garages  Yes No √ 46% 54% √ 17% 83% √ 52% 48% Type of house single  multifamily √ 64% 36% √ 100% √ 100% Air Conditioning Yes  No √ 40% 60% √ 100% √ 81% 11% Frequency of AC use Frequently  Sometimes Don’t know Never √ 21% 19.5% 59.5% 0 √ 42% 46% 12% √ 15%  7% 78%  56  For homes that were part of the CHILD study, home information was gathered from both a questionnaire completed by the parents and the home inspection conducted by research technicians. For homes that were  part of TCHEQ, a large amount of  housing characteristics data were  also available from a questionnaire that was administered at study baseline (633 questions). Finally, from the WOEAS homes for which information on a wide range of housing characteristics and time-activity patterns was collected twice, we used the baseline questionnaire. The questionnaires included questions that were unique to each cohort as well as other questions common across all studies (see Table 2.3), which were recoded to generate a set of harmonized data. Harmonized variables included data related to the season (defined using  heating degree day) based on the date when  samples were collected, the type of floor (recoded as smooth for hard wood, vinyl and other smooth surfaces, carpets for rugs and carpets, and mixed for samples collected from both smooth and carpet surfaces), 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 use of air conditioning (recoded as never, sometimes and frequently), the cleaning frequency (recoded as rarely, sometimes and frequently), and the usage of windows (recoded and grouped from different questions in the CHILD questionnaire) coded into 5 categories: usually open/sheer; covered with blinds or curtains; sealed; open daytime/covered nighttime; other.  2.3 Statistical analysis We first analyzed the association between the mixture of hopanes in outdoor air and indoor dust by comparing the relative abundance of the most abundant monomer. Specifically, we calculated the ratio of the concentration of 17α(H), 21β(H)-Hopane to the total concentration of the eleven monomers. We then compared this relative abundance between the outdoor air and indoor dust samples in each city, and examined this association after accounting for temperature and evaluating multicollinearity between predictors (assessed by the variance inflation factor) in linear regression models . 57  After examining the distribution of hopane concentrations in a pooled analysis of dust samples from all cities (hereafter “pooled analysis”) and separately within each city (“city-specific analysis”), we applied a log transformation to the total hopane concentration (i.e. sum of the 11 monomers) distribution across all cities and within each city. Only hopane concentrations in the family room were considered in the analysis since homes from WOEAS and TCHEQ studies did not provide samples from the bedroom and the bedroom samples in CHILD included a mixture of floor and bed samples. Prior to examining the association between total hopane concentrations and GIS variables in the pooled analysis, we fit a varying intercept model to assess the between- and within-city variability. Both in the pooled and city-specific analyses, questions on lifestyle factors and housing characteristics were examined in bivariate analysis as potential confounders or effect modifiers for the hopane – geographic predictor relationships.  The same model building approach described  by Henderson et al. (Henderson et al. 2007) to generate physically meaningful predictive models was adopted and consisted of the following steps:  (1) Rank all variables by the absolute strength of their correlation with the hopane concentration; (2) Within each sub-category (e.g. all buffer sizes for highway lengths),  keep only the highest-ranking; (3) to avoid collinearity examine the correlations between all GIS predictors retained from step 2 as well as questionnaire variables using 0.6 as a cut-off value ; (4)enter all remaining variables into a stepwise linear regression; (5) remove from the available pool any variables that have insignificant t-statistics and variables that show a direction of effect opposite of a priori hypotheses. Steps 4 and 5 were repeated until a parsimonious final model was obtained.  The study methodology was reviewed and approved by both the University of British Columbia Behavioral Research Ethics Board (ethics certificate number: H11-03231) and the Clinical Research Ethics Board (H07-03120). 58  2.4 Results 2.4.1 Outdoor vs. indoor hopane concentrations comparison All samples were above the GC/MS limit of detection (LOD). In all cities the monomer 17α(H), 21β(H)-Hopane was consistently detected and showed the highest abundance in the suite of analyzed compounds, therefore the comparison of outdoor and indoor hopane ratios was performed using this monomer relative to the sum of all monomers. The sampling from the air monitoring stations was conducted in 2010 at fixed time points which result in concentrations from air samples with a discrete distribution (Figure 2.1) compared with the house dust samples which were collected throughout the year. In air, the range of the 17α(H), 21β(H)-Hopane relative abundance (0.2 to 0.4) generally corresponded to the same relative abundance in house dust. The correlation of the 17α(H), 21β(H)-Hopane relative abundance in outdoor air and house dust was moderately strong, yet significant (r=0.48, p<0.05).   Figure 2.1: Association between outdoor air and house dust hopanes major monomer ((17α(H), 21β(H)-Hopane) relative abundance.    59  After excluding an outlier sample (see in Figure 2.1 data point near zero where the ratio in 17α(H), 21β(H)-Hopane was depleted due to a very low concentration in all monomers), we also examined the relation between the outdoor and the indoor relative abundance  in linear regression accounting for the effect of season, and found a stronger statistically significant relationship (slope=0.92, t=6.1) compared to the association without adjustment for season (slope=0.72, t=5.9). The correlation was still significant when the outlier was included. The effect of season was stronger during the spring and summer (r>0.5) than during the fall and winter (r<0.5).  2.4.2 Pooled results Hopane levels in individual homes varied from a low of 0.4 ng/mg of dust in a Toronto home to a high of  41.8 ng/mg of dust in a Vancouver sample, after excluding an outlier in Toronto from the CHILD study where the concentration was 160.3 ng/mg (more than 29 times higher than the median); for this home we examined the land use characteristics, the road network, and potential outdoor sources as indicated in the questionnaire and did not find any difference that would explain such a high concentration. Analyses were thus run with and without this sample (Table 2.4).  Windsor had the lowest overall indoor hopane levels with a mean level of 5.8 ng/mg (GM= 5.1 ng/mg, GSD =1.8) while the sample of homes in Vancouver showed the highest mean concentration of 9.3ng/mg (GM= 6 ng/mg, GSD= 2.9) when the high (Toronto) outlier was excluded (if the Toronto outlier was retained, then Toronto was ranked first with an AM=9.9ng/mg).  After fitting a null random effect model, the intraclass correlation of 0.007 indicated that city clustering would not contribute to explaining the variability in total hopane concentrations. We therefore built a model without using a city-specific random intercept in the regression analysis and all samples were treated as independent.  60  From the harmonized questions, only relative elevation and heating degree days at the time of dust collection showed a statistically significant relationship with hopane concentrations. Distance to highway (DIST_RD1) had a statistically significant association with hopane concentrations, but its direction of effect was opposite to what we expected and was therefore excluded from the model. The final model with relative elevation and heating degree days as predictors explained only 6% of the total variability in the hopanes concentration. Including distance to highway did not appreciably improve the amount of explained variability (adjusted R2 =0.08). Excluding the high outlier home in Toronto led to a model with the addition of the presence of an AC unit in the home along with the same predictors as above, but with less overall variability explained (adjusted R2= 0.04).  2.4.3 City-specific modeling results Given the availability of LUR models for predicting NO2 in each study area, we extracted the NO2 concentration at the geocoded participants’ home addresses and examined the correlation of hopane concentrations in house dust with city-specific LUR NO2 estimates in each city. Results (Table 2.5) indicated no statistically significant associations except in Windsor (r=0.44, p<0.05) and Edmonton (r=0.58, p<0.05). Leveraging the availability of city-specific LUR models, we further examined separately for each city the association between hopane concentrations and the variables that were used both in the city-specific LUR models describing the NO2 levels (Table 2.2) and those that we generated for the pooled analysis (Table 2.1). The amount of variability explained in each city varied from 10% in Vancouver to 80% in Edmonton (Table 2.4).     61  Table 2.4: City-specific determinants of hopane concentrations in house-settled dust. City Final model  with regression coefficients Partial R2 Model Adj.R2 Edmonton log (hopanes) =  3 – 0.13 cleaning frequency  - 1.5 Smooth Flooring -0.15 Air Conditioning 0.78 0.78 0.35 0.80 Toronto Log (hopanes) = 2.69 -1.03 Smooth Flooring -0.008 Elevation + 0.88 Attached Garage -0.66 Detached Garage 0.29 0.13 0.13%* 0.45 Windsor** log (hopanes) =  5.6 +  0.5 elevation + 0.17 RD1_100 0.36 0.13 0.39 Winnipeg log (hopanes) = 1.45 – 0.057Heating degree days  – 1.33 multifamily house 0.17 0.16 0.33 Vancouver log (hopanes) =  1.9 – 0.09 Heating Degree Days  – 0.07 Shoe removal 0.09 0.07 0.10 * The ‘detached garage’ variable has three categories: No garage; Attached garage; Detached garage. ** In Windsor, elevation and distance to the Ambassador Bridge were strongly and significantly correlated. An alternative model with Distance to Ambassador Bridge yielded similar results, yet with smaller R2  Overall, in each city the determinants of indoor dust hopanes were predominantly related to home-specific factors (cleaning, use of AC, shoe removal) and meteorology, except for Windsor where the final model included the length of major roads in a 100m buffer (Table 2.3). In Toronto, the spatial variability provided by the TCHEQ samples was very limited as all homes were within a restricted geographic area within the city. Hence, an additional sub-analysis was run for Toronto with only the CHILD homes 62  included. This model (not shown) did retain GIS variables (open area within 1000m buffer and elevation) as well as variables related to other possible sources of hopane emissions (garage type, presence of a construction site within 100m) and finally home-specific factors (i.e. the type of floor surface) and explained 86% of the overall variability in indoor dust hopanes. After excluding the house with the outlier concentration value, however, the final model, with an R2 = 0.3, had exactly the same predictors as those shown in table 2.3 where samples from both the TCHEQ and CHILD study homes in Toronto were included.  While the association of hopanes indoors in relation to GIS variables typically used as surrogates for TRAP was only modeled for samples collected in living rooms, Table 2.5 shows the concentration in each city by room type and the number of homes (from the CHILD study) where two rooms were sampled. In CHILD participating households dust samples from the living room as well as a second composite sample from subject child’s mattress and adjacent flooring. The ranking by decile showed that the hopanes concentration in the living rooms was significantly greater than that found in the bedrooms.   63  Table 2.5: Geometric Mean (GM) and Geometric Standard Deviation (GSD) of total hopanes concentrations in the study population, by room, by homes and correlation with city-specific modeled NO2 City Homes N (number of homes) Hopanes concentration (ng/mg) Pearson correlation   Bedroom  Family Room Average Between NO2 and family room  n GM(GSD)  n GM(GSD)  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.)  n: total number of samples; n.s.: not statistically significant association.64   2.5 Discussion   Assessing indoor levels of TRAP through the collection and analysis of settled house dust is a new area of study and has the potential to reduce exposure misclassification and increase specificity. In this investigation, we compared measured dust hopanes with ambient air hopane measurements and with GIS-derived land use variables for each city. The availability of contemporaneous cohort studies (CHILD, TCHEQ and WOEAS) offered a unique opportunity to gather a sample of 171 homes where dust was collected using similar protocols in 151 living rooms and where hopanes were analyzed by tandem GC/MS at the same laboratory using a standardized protocol. This sample was representative of different settings ranging from highly urban locations such as Toronto to smaller and less urban cities such as Winnipeg, while also including major transit hubs such as Windsor where an important Canadian-American truck crossing exists. Furthermore, all the cities had previously developed LUR models which reasonably predicted traffic related NO2 spatial variability (from 66% in Vancouver to 81% in Edmonton(Allen et al. 2011; Henderson et al. 2007; Wheeler et al. 2008).  We demonstrated that hopanes can be consistently detected in house dust samples regardless of the type of city and the dust collection location. In addition, after controlling for heating degree days and its impact on infiltration, the major hopane monomer relative abundance in house dust and outdoor air samples were modestly but significantly correlated (r = 0.48), suggesting similar hopane sources in the two samples, but there remains substantial unexplained variability in indoor levels. This comparison had relatively good external validity given that the ambient monitoring sites were located to capture urban background concentrations rather than hot spots and since samples were collected in and matched for all seasons. This correlation was stronger in the summer compared to the winter, suggesting an impact of infiltration as windows are more likely to be opened on warmer days. Since hopanes in house dust 65  accumulate over relatively longer periods of time compared with hopanes in air samples and may have undergone many changes and cycles in temperature, it is likely that the seasonal effect shown in the literature (Olson et al. 2008; Olson and McDow 2009; Rutter et al. 2011)  may not hold in this context. In addition, dust sampling, which often is a readily available matrix for sampling multiple agents in epidemiological studies, including hopanes as demonstrated in this study, does not represent similar constraints (e.g. logistics) as those imposed by particle infiltration measurements. Examining associations between hopanes concentrations and geographic predictors in a pooled analysis indicated that only a small degree of variability in hopanes concentrations in dust was explained by the final model. Further, in this analysis, higher levels of road variables were linked to lower levels of hopanes. Despite the advantages of pooling data from different cohorts, this effort was hindered by the absence of consistency in the supplementary data collected via questionnaires since each study used its own set of questions. While we inspected each question and the research technicians’ notes for each sample of house dust collected in order to generate harmonized variables that could affect the hopanes concentration in house dust, recoding variables may have resulted in a loss of specificity. Unlike the pooled analysis, the city-specific analysis provided more insight into the utility of hopanes as possible markers for TRAP as a moderate to large amount of variability in the total hopane concentration in house dust was explained in each model. This analysis, however, was hindered by the lack of consistency between cities in terms of main predictors of indoor hopane concentrations.  We examined potential modifiers that could alter the relationship between LUR variables and hopanes in dust for each city separately. In addition to geographic surrogates of TRAP, all cities had at least one predictor of hopanes concentration related to the indoor environment or home construction. Thus, the variability in settled dust hopanes concentrations appears to be a function of a mix of parameters that are not exclusively related to traffic emissions.   66  For example, indoor hopanes in settled dust may also result from coarse PM being tracked indoors. A recent analysis of indoor PAHs indicated the potential importance of this pathway even after adjusting for carpeting, frequency of vacuuming and indoor burning (Mahler et al. 2010). We examined the association in the city-specific analysis for all CHILD participating homes between shoe removal habits and hopane concentrations. We found that only Vancouver samples were correlated with shoe removal habits in the expected direction. Collection of supplementary field data remains a crucial component for assessing the utility of hopanes in house dust since tracked dust seems to contribute to hopanes concentration in house dust. This information was only available in the CHILD homes, and could therefore not be assessed in the pooled analysis.  In our investigation we made a critical assumption that hopanes have few sources beyond engine oil lubricants as we were not able to find information on indoor hopanes sources in the literature. Since hopanes are widespread in recent and ancient sediments, they are constituents of all mineral oil or petroleum-based lubricants and it is therefore possible that unaccounted for indoor sources were present.  House dust remains an attractive metric for exposure assessment because it offers a matrix for multiple indoor contaminants, both biological and chemical and both indoor and outdoor in origin, and can be stored for long time periods, thus providing the opportunity to examine additional research questions when necessary. The utility of hopanes in house dust as an indicator of infiltrated TRAP is limited in the absence of better understanding of its deposition and stability in house dust. House dust is heterogeneous matrix with a complex history in each home as it accumulates contributions from multiple sources including not only fresh emissions of combustion-related particles but also road dust which also contains hopanes. The mode of accumulation also contributes to the variability of vacuum dust. Several factors that may vary among study participants can affect the concentrations of hopanes: cleaning practices and sampling surfaces (carpeted vs. non-carpeted) play a role in the amount of chemicals that deposit inside 67  the homes as shown in the city-specific analysis. In addition, the metric of exposure for hopanes still lacks consensus as hopanes can be measured in terms of loading of chemicals (concentration normalized by surface area sampled) or expressed as the more traditional approach of normalized concentration to mass of dust collected. Differences in the choice of metric would relate mostly to cleaning practices, which we have tried to account for in our investigation. Future investigations of other species, such as PAHs, on their own or in combination with hopanes, may offer additional insight into the utility of settled house dust as a surrogate for TRAP exposure.  In our study, we compiled the information about presence and frequency of use of air conditioning as this has been shown to be an important predictor of PM infiltration(Allen et al. 2012a), but we found limited explanatory power in both pooled and city-specific analysis. PM infiltration varies with particle size, with a maximum infiltration efficiency for diameters of approximately 0.2-0.3 µm (Sarnat et al. 2006), while the size distribution of hopanes ranges between 0.7 and 3.3 µm (Huang et al. 2012) which would imply that hopanes infiltration efficiency may be low and might therefore explain variability in the outdoor/indoor correlation (Olson et al. 2008). We could expect that in presence of higher concentration level of hopanes in ambient air (i.e. better ability to detect hopane monomers), the analysis of relative abundance in ambient and indoor hopanes would have shown less unexplained variability.  Overall, the results indicate the indoor dust hopanes concentrations depend on both outdoor TRAP concentrations and a variety of home-specific variables such as cleaning, floor type, and presence of AC. This conclusion is supported by an additional analysis where we examined the relative variation explained by LUR NO2 compared to home-specific factors and also found that in some cities the correlation between hopanes and LUR NO2 is only revealed when accounting for variation due to such home-specific factors.  68  We examined the utility of measurements of hopanes in house dust as exposure indicators for infiltrated, time-integrated, traffic-related pollutants. When combined with behavioral factors retrieved from questionnaires, and geographic determinants, hopanes in house dust may have the potential to be used as surrogates for infiltrated TRAP. Further characterization of the determinants of hopanes in house dust may result in an improved exposure measure for epidemiologic studies to more precisely analyze relationships between TRAP and chronic health effects.  69  Chapter 3: Early-life exposure to traffic-related air pollution and development of atopy in the CHILD birth cohort study8  3.1 Introduction The incidence of allergic diseases has increased sharply, especially for people living in urban areas, which raises an important public health concern given the trend to urbanization worldwide (D’Amato et al. 2010). While associations between exposure to air pollution derived from traffic emissions and allergic exacerbations have been demonstrated, the potential role of traffic related air pollution (TRAP) in the onset of allergic diseases is uncertain (Heinrich and Wichmann 2004).  Several epidemiological studies have reported associations between atopy or other allergic phenotypes and exposure to some TRAP exposure surrogates, including nitrogen dioxide (NO2) (Gruzieva et al. 2012; Janssen et al. 2003;  Brauer et al. 2007; Morgenstern et al. 2008). However, a number of other studies have not observed these positive associations (Gehring et al. 2010; Oftedal et al. 2007). Differences in TRAP exposure assessment approaches are one possible explanation for these divergent findings. Despite the importance of early life exposures in the development of allergy, information on the effect of air pollution exposure, particularly during pregnancy, on allergic responses early in life has rarely been assessed (Aguilera et al. 2013; Esplugues et al. 2011; Mortimer et al. 2008) and birth cohort studies addressing this relationship are rare (Bråbäck and Forsberg 2009). Exposure assessment  methods  have evolved from self-reported measures (e.g. proximity) (Janssen et al. 2003) to atmospheric dispersion (Oftedal et al. 2007) and land use regression (LUR) models capturing within-city air pollution variations (Brauer et al. 2007b; Gehring et al. 2010b). Most LUR models consider annual average concentrations                                                      8 A version of this manuscript has been published: Hind Sbihi, Ryan W. Allen, Allan Becker, Jeffrey R. Brook, Piush Mandhane, James A. Scott, Malcolm R. Sears, Padmaja Subbarao, Tim K. Takaro, Stuart E. Turvey, and Michael Brauer. 2015. Perinatal Exposure to Traffic-Related Air Pollution and Atopy at 1 Year of Age in a Multi-Center Canadian Birth Cohort Study. Environmental Health Perspectives 2015, DOI:10.1289/ehp.1408700 70  which may over or under-estimate personal exposures as traffic markers such as NO2 are highly variable in time and space (Mölter et al. 2010). Recently the temporal specificity of LUR models has been improved by applying forward (Morgenstern et al. 2008) or backward (Brauer et al. 2007b; Gehring et al. 2010b) time trends from fixed site monitoring stations to the LUR predictions. However, given the potential importance of specific short-duration periods of development, refined approaches to capture fine-scale seasonal variability are needed (Mortimer et al. 2008). Exposure assessment is further complicated by the mobility of study participants, a consideration  seldom addressed and rarely measured in large cohort studies (Arrandale et al. 2010), yet one with  considerable influence on personal exposures (Nethery et al. 2008a). The inclusion of exposure in other locations where children spend a significant amount of time, such as daycare facilities, is likely to increase the precision of the exposure estimates. Finally, all prior studies have used ambient concentrations at a given location to estimate exposure, although most time is usually spent indoors, especially in the early years of life (Leech et al. 2002). Considering that indoor infiltration for some pollutants can vary significantly between residences (Clark et al. 2010) substantial exposure misclassification may be present in prior studies.  To address these limitations we examined the association between exposure to TRAP and the development of atopy in a population-based multi-center birth cohort, the Canadian Healthy Infant Longitudinal Development (CHILD) study. We estimated exposure to NO2 as surrogate for TRAP during pregnancy and the first year of life while accounting for both spatial and temporal variability in ambient concentrations. In stratified analyses we evaluated the influence of time-location patterns, daycare attendance, and modeled estimates of infiltration.    71  3.2 Methods 3.2.1 Study population  CHILD is a prospective longitudinal national birth-cohort that recruited over 3,600 families between 2008 and 2012 in four Canadian cities (Vancouver (2.31 Million inhabitants), Edmonton (1.16 M), Winnipeg (0.73 M), and Toronto (5.58 M)) (http://www.canadianchildstudy.ca) (Takaro et al., 2015). Participants had to be pregnant, 18 years or older (19 in Vancouver), reside in reasonable proximity to a recruitment center (originally set about 50km distance from the study center), communicate in English, provide informed consent, intend to give birth at one of the recruitment centers, and able to provide valid personal contact information as well as two alternate contact individuals. Eligible infants had to be born at ≥ 35½ weeks gestation with weight ≥2500 grams. Infants were excluded under any of the following criteria: conception via in-vitro fertilization, product of a multiple gestation pregnancy, major congenital abnormalities, spending less than 80% of nights in the principal residential home. Recruitment ideally occurred at or soon after the routine 18 week ultrasound examination, but a significant proportion was recruited in later pregnancy (34% between 24 and 30 weeks, and 31% after 30 weeks of gestation). Each study center (universities and hospitals) obtained ethics approval from their governing health and ethics board and the CHILD study was reviewed and approved by the Hamilton Integrated Ethics Board (certificate number: 07-2929). The start of follow up was defined as the date of conception based on the expected and actual dates of delivery. End of follow up was defined as the earliest of October 15th 2013 or when the participating child was assessed with skin allergy testing at approximately one year of age.  3.2.2 Skin prick tests To determine individual allergen sensitization, epicutaneous skin tests were administered to each infant at approximately 1 year of age for the six inhalant (Alternaria alternata, Cat Hair, Dog Epithelium, House 72  Dust Mites (Dermatophagoides pteronyssinus and D. farinae), German cockroach) and four food (whole cow’s milk, egg white, soybean, peanut) allergens. Histamine 1mg/mL and Glycerin were used as positive and negative controls. To define maternal atopy, each mother was tested with a panel of allergens:  Alternaria alternata, Cat Hair, Dog Epithelium, House Dust Mites, German cockroach, Cladosporium sp, Penicillium mixed, Aspergillus fumigatus, Midwest trees, grass mix, weeds, mixed ragweed and peanut. Atopic status was determined using a positive response to any allergen. We defined two clusters of atopic responses in the children – inhalant allergy and food allergy.  Participants reporting use of antihistamines during the seven days prior to the date of skin prick test were excluded. The wheal responses were measured at 10 (histamine) and 15 (allergen) minutes. We averaged the maximum diameter and its orthogonal, and defined a positive response as a wheal diameter ≥2mm greater than the response to the negative control. We included all participants with a positive response to histamine and no response to glycerin, or those with one or more positive responses ( 2mm) to any allergen, even if there was a weak or no response to histamine. Participants with a positive response to one or more allergens but also a response to the negative control were included with adjustment for the negative control (subtraction of the mean wheal diameter of the negative control from each positive test wheal diameter). In some cases specific tests were omitted (e.g. some families declined infant peanut testing) and these specific allergens responses were recorded as missing but all other data from that participant were included. Participants with no response to histamine and no response to any allergen were excluded, as were participants with dermatographism, when the response to the negative control was as large as any other response.  3.2.3 Air pollution exposure assessment City-specific LUR models were developed to estimate NO2 concentrations and used to assign exposures at the residential locations of all participants. The models are described in detail elsewhere and were 73  developed at different time points between 2003 and 2008 (Allen et al. 2011; Henderson et al. 2007; Jerrett et al. 2007). Models were developed using road and land use data provided by Desktop Mapping Technologies Inc. (DMTI http://www.dmtispatial.com), as well as additional city-specific data sources (e.g. traffic counts from the city of Toronto).   While individual models differed between locations, the variables used for model building had relatively similar grouping categories (Table 1.4): (1) land use; (2) road and traffic; (3) population; (4) physical geography; and (5) meteorology. All statistical models were consistently developed using the methodology adopted in the initial Vancouver model (Henderson et al. 2007). These models allow cost-effective, relatively precise individual predictions of TRAP at the participants’ residence. Individual estimates of exposure were extracted for each geo-referenced residence reported by participants from address at conception to the address where the participating child resided at age one. Exposure to NO2 was assessed in two case scenarios. In the first scenario, we used the traditional approach of using only the address at the time of birth. Secondly, for those reporting multiple addresses we considered their exposure at every address during the study period and computed their time weighted average exposure. All exposures were subsequently temporally adjusted based on local fixed-site ambient monitoring data on a bi-weekly basis (see equation 1 for  one 2-week interval) over the entire pregnancy and first year of life  by multiplying the LUR estimate by the ratio of the contemporaneous average concentrations measured at all fixed-site governmental monitoring stations (Vancouver: N=12, Toronto: N=7,  Edmonton: N=4, and Winnipeg N=2) to the annual average at the same stations for the year the model was developed.  adj9. NO2 = NO2(LUR) (2-week NO2(monitor)/yearly NO2(monitor))  [1];                                                       9 adj.NO2 refers to temporally adjusted individual estimate of nitrogen dioxide. 74  Since the LUR models did not in all cases cover the full areas where participants resided with the exception of Vancouver, it was necessary to impute exposures for those outside of the LUR model domain. We restricted the locations where exposures were imputed to those residences within 50 km from the municipal center, defined as the location of the city hall. Participants with home addresses located outside of a 50km buffer from the city hall were not assigned TRAP exposure. To reduce discontinuities in exposure estimates on the periphery of the study area, we first defined core city limits within the original LUR surface where the models were applied directly to estimate exposures. For all homes outside these core city limits and within the 50km buffer, the minimum value of the LUR surface in the core city area was assigned to homes within 100m of highway (defined by standard road classification categories (DMTI Spatial Inc., Markham, Ontario), with categories 1 (expressway), 2 (principal highway), and 3 (secondary highway)) and the minimum value of the original LUR surface was assigned to homes located farther than 100m away from a highway. Finally, participants with a TRAP concentration estimated for at least 75% of the time window of interest (i.e. full pregnancy or first year of life) were assigned the mean exposure of the corresponding time window. 3.2.3.1 Time-activity patterns and daycare attendance The CHILD home environmental questionnaires were used to gather information on the micro-environments within and outside the home of each participating child since birth. From the 3, 6, and 12 month home questionnaires we derived: (a) the average time spent away from home (i.e. time weighted average of hours per day spent outside the house based on typical weekday and weekend time-activity patterns) that was divided into two strata by the city-specific median time away from the house, (b) whether the participating child ever attended a daycare or other indoor location for at least 7hrs per week or 1 hr. per day regularly at any time point during the first year of life. Using these estimates we evaluated the association between TRAP exposure and atopy in stratified analyses comparing children ever/never 75  attending daycare and children spending more time or less time than the city-specific median outside the home.  3.2.3.2 Infiltration To assess the potential impact of ventilation, we adapted a model previously developed (Allen et al. 2012a) to predict fine particle (PM2.5) infiltration in 6 US cities as part of the MESA Air study. The model included information on residence characteristics and behaviors related to infiltration, including presence/use of air conditioning (AC), outdoor air temperature, window opening, and use/type of heating. As the CHILD and MESA-Air questionnaires differed, we mapped the MESA-Air variables onto those collected in CHILD to predict infiltration (Table 3.1). For behaviors that varied seasonally, participants were asked about typical behavior in mid-summer and mid-winter. Season was defined by the Environment Canada daily temperature data (http://climate.weather.gc.ca/), collected from the same monitors from 2008 to 2013, where the warm season was defined as every two-week period with a median temperature above 18°C, similar to the MESA Air study. We predicted home-specific seasonal infiltration efficiencies and classified homes based on the city-specific 80th percentile of infiltration efficiency for each season, where participating houses above this threshold were classified as “leaky”. As our exposure of interest was NO2, we used modeled particle infiltration as a surrogate for home ventilation and infiltration of TRAP in general, and assessed effect modification by infiltration in stratified analyses.      76  Table 3.1: Particle Infiltration (Finf) variables; mapping questions used in the MESA-Air (Allen et al. 2012) and CHILD cohorts. Season MESA-Air Variable CHILD Variable  Partial R2 in the MESA-Air model Warm (≥18° C) Central AC used > ½ time in past July Central AC used regularly in the summer a 0.560 Windows open ≥ ½ time in past summer Windows open ≥  1hr  more than 2 times/week in mid-summer a 0.080 Central AC used > ½ time in past July and 2-week avg. outdoor temperature > 230C Central AC used regularly in the summer a and 2-week government monitors averageb 0.051 Central AC used a few days in past July Central AC use occasionally in the summer a 0.013 2-week avg. outdoor temperature > 230C 2-week avg. outdoor temperature   > 230C  b 0.000 Cold (<18° C) 2-week avg. outdoor temperature (0C) 2-week avg. outdoor temperature b 0.222 Home has forced air heat Home has furnace c 0.166 Windows open ≥ ½ time in past summer Windows open ≥  1hr  more than 2 times/week in mid-summer c 0.069 a:Home environmental questionnaire; b: Environment Canada (http://climate.weather.gc.ca/), c: Home inspection;   3.3 Questionnaires and home inspections Covariates relating to (1) indoor and outdoor environments, (2) socioeconomic and parental risk factors, and (3) nutrition were derived from self-reported questionnaires, or from inspections by trained technicians (for timetable of assessments, see Table 1.2). Using standardized protocols, technicians carried out a home assessment when infants were ~3-4 months old and gathered information about residential characteristics and activity patterns of the occupants. Secondhand smoke exposure (SHS), 77  presence of leaks, mold, pets, insects and other pests were assessed by questionnaires at four time points (at time of enrollment, and at 3, 6, and 12 months of age).  Three measures of SES were examined using the father and mother’s education status and the household income reported at enrolment.  Information on maternal history of asthma, smoking (before/during pregnancy), secondhand smoke exposure during pregnancy and prior live births was obtained at enrolment by questionnaire. Additionally, birth chart data provided information on type of delivery (vaginal vs. Cesarean-section), length of gestation, and child’s sex. Information on breastfeeding/feeding practices was provided by the mother in postnatal questionnaires (approximately at 3, 6 and 12 months). Breastfed participants were defined as having been breastfed when the mother answered positively at any time point. Children with formula, soy and cow’s milk intake were similarly derived. Variables relating to the introduction of solid foods were derived using any positive answer to the same question across all three questionnaires asking mothers to “indicate which foods you currently feed your child”. Timing of food introduction was not considered in this investigation. 3.4 Statistical analysis Prior to investigating the association between TRAP and atopy, we assessed bivariate associations between sensitization at one year and all covariates mentioned above from the environmental (furry pets, SHS, leaks, mold, pests), parental (parity, maternal asthma, maternal atopic status, maternal smoking status during pregnancy, type of delivery), socio-economic (income, maternal and paternal education) and diet risk factors (breastfeeding, formula, soymilk,  and introduction of different solid foods) for each of the three outcomes of interest (sensitization to any allergens, to any inhalant allergens and to any food allergens) as well as sex and presence of an attached garage, to test for potential confounding. In these analyses, we first examined covariates for each time point separately. Subsequently, never/ever variables were generated whereby for any positive response to a risk factor, the participant was categorized as 78  having this risk factor. Further bivariate analyses, evaluating the associations with each outcome separately (sensitization to any allergens, to any inhalant allergen, and to any food allergen), were carried out with the latter never/ever variables. Covariates that were significant predictors (p<0.05) of each of the three outcomes of interest from the bivariate analyses were considered in models of NO2 and atopy for each time window. We performed a manual stepwise backward multiple variable logistic regression with city as random intercept to obtain a parsimonious model with significantly (p<0.05) associated predictors of each outcome.  In the pregnancy time period, the model for inhalant allergens included presence of an attached garage and mold, whereas the food allergens model included maternal atopy, presence of furry pets, and household income. In the first year of life, inhalant models controlled for presence of furry pets and any consumption of nuts, while the food allergen model controlled for maternal atopy, presence of furry pets, and any consumption of eggs, processed cereals, and peanuts. For sensitization to any allergen, the significant predictors were similar to those obtained for the food allergen model in the first year analysis. During pregnancy, the model for this outcome included maternal atopy and presence of furry pets.  The main analysis, which included a random study center intercept to adjust for the clustering within city, focused on predicted outdoor NO2 exposure at the home address(es) only, while we assessed time spent away from home, daycare attendance and infiltration in sensitivity analyses. Additional analyses for effect modification by parity, sex and maternal atopy were carried out. Predictors for these analyses were the same as those used in the main analysis. Effect estimates are presented for a 10 µg/m3 increase in NO2, approximately representing the overall standard deviation.   79  3.5 Results 3.5.1 Population characteristics and atopy  Atopy data were examined for 2482 children who had been assessed at age one by October 15th, 2013. Five children were excluded from the analysis due to non-interpretable skin prick test results (n=2), antihistamine medication taken prior to skin testing (n=1), or test postponement (n=2). Of the 2477 participants, 400 infants were sensitized to at least one of the 10 tested allergens (Table 3.2). Table 3.2: Atopic outcomes (inhalant, food and any allergies) by city among 2477 CHILD participants with valid skin prick test at age 1 City (N, % sensitized) Inhalant sensitization n(%)a Food sensitization n(%)a Any sensitization n(%)a Vancouver (575, 23% ) 55 (42) 92 (30) 132 (33) Edmonton (641, 17%) 28 (21) 85 (28) 108 (27) Winnipeg (680, 9%) 13 (10) 50 (16) 60 (15) Toronto (581, 17%) 36 (27) 82 (27) 100 (25) Total (2477, 16%) 132 (5) 309 (12) 400 (16) aData are shown as n, number of sensitized children and (%)proportion within each atopy outcome examined.   Across all cities, 309 (12.5%) infants were sensitized to any food allergen while 132 (5.3%) had a positive response to any inhalant allergens. Vancouver had the largest proportion of atopic children (23.5%) followed by Toronto and Edmonton (both 17%), and only 9% in Winnipeg (Table 3.2). As expected for infants, prevalence of individual allergens was low (Table 3.3), precluding the investigation of associations between NO2 and individual allergens. However, in preliminary bivariate analysis, dog (OR = 1.20; 95% CI: 1.01 – 1.21), Der p (OR = 1.14, 95% CI: 1.07 – 1.27) and peanut sensitization (OR = 1.07; 95% CI: 1.03 – 1.11) were individually associated with temporally adjusted exposure at the birth address.   80   Table 3.3: Proportion of positive responses to individual allergen skin prick tests by CHILD city (in (%)) City Alt Der p Der f Cat Dog Blatt Peanut Milk Eggs Soy Vancouver (N = 575) 1.9 1.4 3.0 3.1 1.0 1.2 6.1 8.5 4.0 1.7 Edmonton (N = 641) 1.1 0.5 0.9 1.4 0.3 0.3 6.2 9.2 2.3 1.1 Winnipeg (N = 680) 0.1 0.1 0.1 1.2 0.1 0.4 3.2 5.3 0.4 0.1 Toronto  (N = 581) 1.2 1.4 0.7 2.2 2.2 1.0 6.5 9.1 1.7 2.1  Only participants with non-missing exposure and health data (pregnancy: n=2123; first year: n=2173) as well as completed questionnaire information about environmental, parental, socio-economic, and nutritional risk factors were considered for analysis. Pregnancy and first year of life subsets used for analysis and the full cohort with sensitization results (n=2477) were virtually identical in the distribution of predictors and disease prevalence (data not shown). We also examined the effect of the exposure imputation, which was performed to include participants residing outside of the LUR surfaces. Rates of allergic sensitization were not statistically different between children assigned LUR estimates (16%) and those who had imputed concentrations (13.5%). Covariates considered for the analysis between TRAP and atopy risk were similarly distributed in the sample with imputed concentrations and that with LUR derived exposure estimates (data not shown).  Most mothers were highly educated (93% completed at least some university or college education or more), nonsmokers during pregnancy (97%), non-asthmatics (77%) yet with a high prevalence of sensitization to at least one of the ten allergens tested (61%) and had a vaginal delivery (65%). About half of mothers had previous live births (47%). The majority of participating families (70%) belonged to higher SES based on household income ≥ $80,000/year (Table 3.4)  81  Table 3.4: Cohort characteristicsb among 2477 children at one year of age with valid skin allergy tests, and crude Odds Ratio (OR) for sensitization to any allergens with 95% Confidence Interval (95% CI) N (%)  Non-atopic Atopic OR   95% CI Personal/Maternal covariates  Sex  Male 1282, (52) 1067, (51) 215, (54) 0.94 (0.76,  1.18) Female 1195, (48) 1010, (49) 185, (46) Maternal atopy statusc  Yes 1509, (61) 1227, (59) 282, (71) 1.68 (1.33,  2.12) No 966, (39) 848, (41) 118, (30) Missing 2 0 2 Maternal asthma status   Yes 522, (23) 435 (23) 87, (23) 1.09 (0.83,  1.42) No 1771, (77) 1485 (77) 286, (77) Missing 184 157 27 Maternal smoking in pregnancy   Yes 74, (3) 66, (3) 8, (2) 0.74 (0.35, 1.57) No 2219, (97) 1854, (97) 365, (98) Missing 184 157 27 Maternal smoking ≥ 1 year  Yes 617, (27) 527, (27) 90, (24) 0.85 (0.65, 1.10) No 1674, (73) 1391, (73) 283, (76) Missing 186 159 27 Parity  Has previous births 1075, (47) 914, (48) 161, (43) 0.85 (0.68, 1.07) No previous births 1219, (53) 1007, (52) 212, (57) Missing  183 156 27 Delivery mode  Vaginal  1481, (65) 1244, (75) 237, (73) 1.05 (0.79, 1.37) C-section 501, (22) 413, (25) 88, (27) Missing 495 420 75 Socio-economic covariates  Maternal education  High School 166, (7) 144, (7) 22, (6) ref College or University 1686, (74) 1415, (74) 271, (73) 0.99 (0.61, 1.59) Postgraduate Education 439, (19) 362, (19) 77, (21) 0.98 (0.58, 1.67) missing 186 156 30 Household income   < 40K 155, (7) 132, (8) 23, (7) ref 40-80K 490, (24) 417, (24) 73, (22) 0.96 (0.58, 1.61) 80 - 150K 928, (45) 777, (44) 151, (46) 0.95 (0.59, 1.55) > 150K 511, (25) 431, (25) 80, (24) 0.83 (0.49, 1.40) Missing 393 320 73 Environmentald covariates  Furry Pets   Yes 1134, (65) 977, (54) 157, (44) 0.72 (0.58, 0.90) No  1048, (22) 846, (46) 202, (56) Missing 295 254 41 Garage yes 751, (65) 623, (41) 128, (45) 1.35 (1.03, 1.79) no 1070, (22) 912, (59) 158, (55) Missing 656 542 114 82  N (%)  Non-atopic Atopic OR   95% CI Personal/Maternal covariates  Introduced food in first yeare  Dairy products   Yes 1944, (95) 1639, (96) 305, (92) 0.51 (0.32, 0.79) No 102, (5) 75, (4) 27, (8) Missing 431 363, 68 Processed cereals  Yes 1661, (83) 1413, (85) 248, (76) 0.58 (0.44, 0.76) No 334, (17) 256, (15) 78, (24) Missing 482 408 74 Eggs Yes 1742, (85) 1478, (86) 264, (80) 0.54 (0.40, 0.73) No 303, (15) 235, (14) 68, (20) Missing 432 364 68 Nuts Yes 625, (31) 1180, (69) 240, (72) 0.72 (0.55, 0.93) No 1420, (69) 533, (31) 92, (28) Missing 432 364 68 Peanuts Yes 941, (46) 816, (48) 125, (38) 0.63 (0.49, 0.79) No 1102, (54) 896, (52) 206, (62) Missing 432 365 69 bData are shown as n (%). The percentages are calculated using the number of observations with known values as the denominator. cPositive skin prick test response to any of the allergens tested. dEnvironmental covariates are based on any self-reported positive response during pregnancy and at 3, 6, or 12 months. eInformation on feeding practices are based on any self-reported positive response at 3, 6, or 12 months.  Compared to atopic participants, non-atopic children were more likely in their first year to consume dairy products, eggs, nuts, peanuts, grains and processed cereals,  to reside in a home with pets , and less likely to have a  garage attached  to their home (Table 3.4). Children of atopic mothers were more likely to be atopic than children of non-atopic mothers (crude OR = 1.68; 95% CI: 1.33 – 2.12).   3.5.2 Exposure levels and association with atopy development Exposure estimates were unavailable for 12% of the 2477 children with skin prick test data, as 173 participants had homes located more than 50km from the city centre and 131 did not provide their residential history since enrolment.  83  Among  the 2173 participants with complete residential histories who resided within the 50km buffer, 252 homes fell outside of each city core limits and were assigned imputed NO2 concentrations. Mean exposure levels differed significantly by city; ranging from 28 µg/m3 in Toronto to 9.9 µg/m3 in Winnipeg (Table 1.3). After applying the bi-weekly temporal adjustment and accounting for residential mobility (83% did not change their address), estimated exposures were lower across all cities due to decreasing ambient concentrations between the development of the LUR model and the date of birth (Brook et al. 2014), with a greater decline for older LUR surfaces. The difference between these estimates is not likely due to addresses change, but mostly to the temporal adjustment from original LUR models to the time of the present of investigation. When examining the pairwise differences of spatiotemporal biweekly means by time window, we found significant differences (p<0.05) between exposure during pregnancy and the first year of life, unlike the non-significant differences obtained with estimates not accounting for residential mobility (p=0.3). Compared to TRAP exposure estimated at the birth address with no temporal adjustment, NO2 estimates that incorporated temporal variability in ambient concentrations increased the magnitude of the effect estimates for the first year of life analysis (any allergens: aOR = 1.10; 95% CI: 0.96 – 1.34), yet without reaching statistical significance (Table 3.5). Further, estimates of effect generally increased when temporally adjusted models further accounted for mobility (Figure 3.1B). However the increased spatial resolution also led to larger confidence intervals around the atopy risk estimates.     84  Table 3.5: Adjusted10 Odds Ratios (aOR) for risk of atopy per 10 µg/m3 increase in NO2 exposures during pregnancy and the first year of life.   All Allergens Food Allergens Inhalant Allergens   aOR 95% CI aOR 95% CI aOR 95% CI Pregnancy     Exposure at birth address 1.06 0.95 – 1.28 1.05 0.87 – 1.34 1.05 0.77 – 1.40 Temporal adjustment(birth address)  1.01 0.81 – 1.16 0.99 0.73 – 1.10 1.16 0.86 – 1.61 Spatio-temporal adjustment  1.02 0.86 – 1.22 1.00 0.77 – 1.61 1.18 0.77 – 1.61 First Year    Exposure at birth address 1.05  0.95 – 1.28 1.08 0.95 – 1.28 1.16  0.91 – 1.40 Temporal adjustment(birth address)  1.10 0.96 - 1.34 1.15 0.95 – 1.40 1.22 0.92 - 1.62   Spatio-temporal adjustment  1.16 1.00- 1.41 1.17 0.95 – 1.47 1.28 0.93 - 1.   During the first year of life, NO2 was associated with sensitization to any allergen tested at one year of age (aOR = 1.16; 95% CI: 1.00 – 1.41) when considering temporally adjusted exposures that also accounted for residential mobility. When examining each group of allergens separately, we also found positive, but non-significant associations (aOR = 1.17; 95% CI: 0.95 – 1.47 for inhalant allergies and aOR = 1.27; 95% CI: 0.93 – 1.76 for food allergies) (Figure 3.1B).                                                       10 Model covariates for each outcome/time window pair are as follows: Any allergies and NO2 during pregnancy: mother’s atopic status, presence of furry pets. Any allergies and NO2 during first year: mother’s atopic status, presence of furry pets, consumption of eggs, consumption of processed cereals, and consumption of peanuts Food allergies and NO2 during pregnancy: mother’s atopic status, presence of furry pets, and household income Food allergies and NO2 during first year: mother’s atopic status, presence of furry pets, consumption of eggs, consumption of processed cereals, and consumption of peanuts  Inhalant allergies and NO2 during pregnancy: presence of an attached garage, presence of mold.  Inhalant allergies and NO2 during first year: presence of furry pets and consumption of nuts. 85   Figure 3.1: Adjusted Odds Ratio for risk of atopy per 10 µg/m3 increase in NO2 exposures (triangle) temporally adjusted at birth address, (circle) temporally adjusted and accounting for residential mobility. (A) during pregnancy: inhalant allergens model controlled for presence of an attached garage and mold (n=1836); food allergens model controlled for mother’s atopic status, presence of furry pets, household income (n=1913); any allergens model (n=2123) controlled for mother’s atopic status and presence of furry pets and (B) during the first year of life: inhalant model (n=2058) controlled for presence of furry pets and any consumption of nuts since birth; food allergen analysis (n=2002)adjusted for mother’s atopic status, presence of furry pets, and any consumption of eggs, processed cereals, and peanuts); any allergen analysis (n=2173) adjusted for mother’s atopic status, presence of furry pets, consumption of eggs, processed cereals and peanuts  In contrast, during pregnancy (Figure 3.1A), effect estimates were null for sensitization to any allergens (aOR = 1.02; 95% CI: 0.86 – 1.22) and for sensitization to food allergens (aOR = 1.00; 95% CI: 0.77 – 1.61). In this time window, the association between exposure and inhalant allergens atopy was non-significant (aOR = 1.18; 95% CI: 0.77 – 1.61).  Analyses of the effect of greater or lesser time spent away from the home indicated improved precision in estimates among children spending more time at the home, and identified a potential source of exposure misclassification (Figure 3.2A). Participants (n= 976) who spent more time away from the home (median: 3.3 hours per day for all cities) generally had slightly smaller effect estimates with larger confidence intervals (any allergens: aOR = 1.16; 95% CI: 0.85 – 1.53) than those spending less time (n=1026) away from their home addresses (aOR =1.22; 95%CI: 1.00 – 1.47). This association was likely driven by the sensitization to inhalant allergens (children spending ≤ city-specific median time away from home aOR 86  =1.61; 95% CI: 1.15; 2.19 vs. those spending more than the city-specific median time away from their homes aOR =1.10; 95% CI: 0.69, 1.68).   Figure 3.2: Adjusted OR of atopy per 10µg/m3 NO2 increase during the first year stratified by: (A) time-activity patterns (defined by the city-specific median hours per day based on the three questionnaires submitted after birth around 3,6, and 12 months) among children spending more time (n=976) and those spending less time (n=1026) away from the home; and (B) daycare facilities attendance among daycare attendees (n=765) and children never attending daycare (n=1236).  Stratifying the cohort by daycare attendance (35% attended daycare) also suggested a source of exposure misclassification, as non-daycare attendees (n= 1236) had 61% increased odds of developing atopy (aOR = 1.61; 95% CI: 1.28 – 2.01) whereas the risk was smaller and not significant in children who did attend daycare (n=765; aOR = 1.05; 95% CI: 0.81 – 1.28) (Figure 3.2B). The sensitization to inhalant allergens showed the largest association with NO2 exposure (aOR = 2.1; 95% CI: 1.40 – 3.17 for children never attending daycare vs. aOR = 1.10; 95% CI: 0.77 – 1.54 for daycare attendees). Given these results, we investigated whether exposures other than NO2 (e.g. contact with other children) may play a role in the sensitization onset. Thus, we ran a stratified analysis by presence of siblings which showed that participants who were in households with other siblings had lower odds of developing sensitization to any 87  allergens (n = 874, aOR = 1.16; 95% CI = 0.91 – 1.54) following exposure to TRAP than those with no siblings (n= 1085, aOR = 1.28; 95% CI = 1.0 – 1.54) (Figure 3.3).  Figure 3.3: Adjusted OR of risk of atopy in CHILD families for a 10 µg/m3 increase in NO2 during the first year of life, stratified by presence of siblings in participants families (group with no siblings, n= 1085; group with sibling, n= 874). All models are adjusted for the same covariates as those used in the main analysis (Figure 1B)   For a 10 µg/m3 increase in NO2 exposure, the odds of sensitization to any allergen for children living in homes with greater ventilation (n= 687) was slightly higher (aOR = 1.22; 95% CI: 0.91 – 1.61) than for children living in tighter homes (n= 824, aOR = 1.10; 95% CI: 0.82 – 1.47) during the heating season, but not in the warm season (Figure 3.4).  88   Figure 3.4: Adjusted Odds Ratio of risk of atopy for 10 µg/m3 increase in NO2 during the first year of life stratified by season (defined using weekly average of 18°C as cutoff to define cold and warm) and by home PM infiltration status (defined based on city-specific 80th percentile predicted household PM infiltration efficiency; “leaky” homes: n= 687; “sealed” homes: n=824). All models are adjusted for the same covariates as in the main analysis (Figure 3.1B)  3.6 Discussion In this prospective multi-centre birth cohort study, exposure to NO2 during the first year of life, but not during pregnancy, was positively associated with atopy at age one year. To our knowledge, this is the first birth cohort study where atopy in relation to traffic-related air pollution was determined in the first year of life (Brauer et al. 2007b; Gruzieva et al. 2012; Nordling et al. 2008). Positive associations between NO2 and specific sensitization to common food, but not inhalant allergens were observed in a subgroup of 700 Dutch children from the PIAMA cohort at age 4 (Brauer et al. 2007). In the Swedish BAMSE cohort, exposure during the first year of life was associated with an increased risk of only pollen sensitization at age 4 (no association with food allergens) (Gruzieva et al. 2012).  89  The ability to refine individual estimates of exposure to TRAP by incorporating temporal changes in air pollution concentrations and in participants’ residential mobility led to larger effect estimates; however the improvement in precision of these effects was negligible and not consistently improved across all three outcomes. It is important to note that the exposure assessment was derived from modeled estimates rather than measurements, increasing error propagation in the estimates used for evaluating the association with atopy outcomes. However, the correlation between the exposures during the entire pregnancy and the first year of life decreased with more specific exposure measures suggesting that refined exposure assessment enables improved differentiation between exposure periods. This finding is supported by a study showing that temporally updated (based on air dispersion model data)  LUR models provide accurate exposure estimates (Mölter et al. 2010).  In light of the recently published European meta-analysis of air pollution with allergic sensitization (Gruzieva et al. 2014),  we explored potential effect modification by sex or maternal atopy. Similar to their results, no effect modification by sex was found. However maternal atopy showed borderline significant effects for exposure during pregnancy and smaller magnitude of effect compared with the main model in the first year (aOR for Maternal atopy ˟NO2 interaction term = 1.04; 95% CI: 0.98 – 1.10 during pregnancy and aOR = 1.04; 95% CI: 1.00 – 1.10 during the first year).  We demonstrated stronger associations between TRAP and atopy in our stratified analyses when daycare attendance and individual time-activity patterns were considered. In particular for exposures during the first year of life, when inhalant allergen sensitization was considered separately, participants for whom exposure misclassification was less likely (i.e. those spending more than the city median time at home, and those who did not attend daycare) had stronger associations. In a small subsample of participants providing daycare addresses (n=235), exposures were not significantly different in homes and in daycares (data not shown), making it unlikely that lower exposures outside the homes would explain reduced effects amongst those attending daycare. While the observed differences in these sub-analyses could also 90  be due to more variability suggestive of a classical exposure error, we explored the possibility that the differences observed were related to exposure other than TRAP. Odds ratios of sensitization to any allergens were lower for participants spending more time in the home or not attending daycare compared to daycare attendees or those spending more time away from the home for the same rate of TRAP increase, suggesting that this latter group might be exposed to other exogenous protective exposure such as presence of other children. The additional stratified analysis by presence of siblings seemed to support the argument that exposure to other children is likely to play a protective role in the development of atopy.  Along with refined exposure assessment modeling, major strengths of our study are the prospective design from early in pregnancy and the objective definition of sensitization. Comparisons between the few birth cohorts examining early life  exposures to traffic pollution are complicated by the various definitions of atopy or allergic sensitization, most often assessed by self-reported symptoms (Bråbäck and Forsberg 2009) which can lead to misclassification of outcomes. In the present study, atopic status was based on objective skin prick tests using a common protocol for all participants. Gathering questionnaire and home inspection data enabled us to collect extensive individual data on known and suspected risk factors about indoor and outdoor environments, and parental health status, as well as detailed dietary information that are seldom acquired in large cohort studies as early as in this investigation. However, the number of missing covariates is a limitation as sample sizes for individual analyses were substantially reduced. Despite the advantage of multiple questionnaires and detailed home assessments, the use of self-reported information on environmental risk factors, which may be biased by parental health status, is a concern. Although the cohort was unselected and the prevalence of parental allergy and current asthma similar to that in the Canadian population, there is a bias towards higher SES compared with the general population, as is often the case with birth cohorts. Further, while this is one of few analyses of TRAP to examine the 91  role of  infiltration, our assessment was limited by the use of a model for particle infiltration developed for cities in the US (Allen et al. 2012) to classify infiltration of TRAP in Canadian homes. We mapped variables in the MESA-Air cohort questionnaires to the most similar questions available in CHILD; however it is likely that we introduced some error in recoding the CHILD variables, and thus in developing infiltration estimates which are already difficult to model based on actual measurements. In addition, the model was developed in US cities spanning a wider north-south geographical area, and consequently developed for a hotter climate, which led to a temperature threshold variable (23°C) that might differ from the cut-off obtained using Canadian data. In the case of the cold season infiltration models, we found the expected differences between homes with participants in the “leaky” homes showing stronger associations between TRAP exposure and sensitization to any allergens only. Lack of sufficient power precluded the identification of differences in this analysis. Future studies should consider an infiltration measurement sub-study to develop a study-specific model. Children at one year of age developed more sensitization to food (12.5%) than  inhalant allergens (5.5%) similar to findings in the European birth cohorts in which participants were older and showed higher prevalence rates of sensitization  (BAMSE with 16% and 15% (Gruzieva et al 2012) and PIAMA with 23.9% and 8.5% (Gehring et al. 2010b) for food and inhalant allergens respectively). However, we observed that exposures during first year of life may contribute differently to the potential load of sensitization. In conclusion, this study demonstrates that in cities with low-levels of ambient traffic-related air pollution,  incorporating different tools (GIS, monitoring data, questionnaires, and home environmental assessment) to account for temporal variation, residential history, and time-location patterns in the estimation of individual-level exposures can help clarify the association between early life  exposure to traffic-related air pollution and the development of allergic sensitization to common inhalant and food allergens. 92    Chapter 4: Asthma development in relation to air pollution and greenness in the BAQS birth cohort: a follow up from birth to age 10.  4.1 Introduction Among pediatric respiratory diseases, asthma is the most common pathology and represents a heavy burden of illness and costs (Ismaila et al. 2013). It is recognized that both environmental and genetic factors play a role in its development (Subbarao et al. 2009). Among the environmental factors, air pollution has attracted interest because of its link to asthma severity, its ubiquitous nature and the possibility of preventive strategies via exposure reduction (Guarnieri and Balmes 2014).  Several qualitative critical reviews have suggested causal associations (Bråbäck and Forsberg 2009; Health Effects Institute 2010) between air pollution, traffic-related air pollution, and incident childhood asthma, although recent meta-analyses of the epidemiological evidence show diverging results. An adverse effect of outdoor air pollution on asthma incidence, but no effect with lifetime period prevalence of asthma, was reported in meta-analyses of studies published up until 2010 (Anderson et al. 2013a, 2013b). Examining air pollution from traffic sources, an American meta-analysis showed adverse effects on childhood asthma (Gasana et al. 2012) while no evidence of an association was found among children followed up to age 10 in a meta-analysis of birth cohorts from the European Study of Cohorts for Air Pollution Effects (ESCAPE) initiative (Mölter et al. 2014b). While the results of these meta-analyses appear contradictory, a possible explanation could be that the increased risk in incidence reflects the unveiling of the presentation of subclinical phenotypes of asthma (Gowers et al. 2012) which highlights the need to examine asthma incidence over different age periods. The current study is able to build on the previous literature by examining the risk of asthma onset among two different age groups (preschool and 93  school-aged) using data from a 10-year follow-up of over 68,000 children as part of the Border Air Quality Study (BAQS) in British Columbia, Canada.  Traffic not only produces air pollutants, it is also likely to affect urban form via reduced access to green spaces. Green spaces have been linked to positive health outcomes (Lee and Maheswaran 2011), to higher physical activity (Roemmich et al. 2006) which is itself associated with decreased asthma development (Sherriff et al. 2009), and to reduced air pollution provided by trees, thereby protecting respiratory health (Lovasi et al. 2008). A recent review on the role of natural spaces in children’s health concluded that reduced access to open spaces is related to higher relative prevalence of asthma and allergic conditions (McCurdy et al. 2010). However, a birth cohort study conducted in New York, reported that asthma at age 7 was significantly increased in those residing in streets with higher tree canopy coverage (Lovasi et al. 2013). Overall, longitudinal follow up studies examining the effect of green spaces on asthma development is scarce and inconsistent. The current investigation will also examine the role of greenness in the development of this disease.  4.2 Methods 4.2.1 Cohort description Using linked administrative data sets originating from the British Columbia (BC) Ministry of Health, the BC Vital Statistics Agency, and BC Perinatal Data Registry, the birth cohort comprised all 1999 to 2002 births  in the greater Vancouver metropolitan region (Brauer et al. 2008). The study protocol was approved by University of British Columbia Research Ethics Board (#H04-80161). Children were followed for 10 years providing they continued to reside in the Metropolitan region of Vancouver based on the Ministry of Health registry data and their mothers also had to be registered as continually residing in the study region during their complete pregnancy. Members were lost to follow-up if they moved outside of the study region or had a gap of over half-a year in residency. From the 73, 387 94  births to mothers with complete information on residential history during pregnancy, children were excluded if they died during follow up (n=101) or were a multiple birth (n=988). The analysis cohort comprised 68,320 singletons with complete covariate information. . 4.2.2 Confounders Birth date, sex, birth weight and gestational length of each child were accessed via Vital Statistics data; and maternal parity, maternal age, intention to breastfeed and maternal smoking during pregnancy were obtained from the BC Perinatal Data Registry. Since, no individual-level data were available for other socio-economic factors, income (quintiles) and maternal level of education (quartiles) were assigned from census data at the Census Dissemination Areas (DA), a resolution which approximates the neighborhood-level with target populations of 400–700 persons  4.2.3 Incident asthma case definition  Records of the utilization of all billed outpatient and inpatient medical services from 1999 to 2009 were obtained from the British Columbia Ministry of Health (http://www.popdata.bc.ca/data). Asthma diagnoses were identified from physician billing records and hospital discharge records for codes 493 and J45 from the 9th and 10th revision of ICD, respectively. Using a validated case definition of asthma (Prosser et al. 2008), children with a minimum of two primary care physician diagnoses in a rolling 12-month period or a minimum of one hospital admission over the study period were identified as asthma cases.  4.2.4 Exposure assessment  All spatially-derived exposure variables as well as residential greenness exposure were assigned to corresponding residential histories throughout pregnancy since all residential postal codes were recorded at each contact with the health care system. An average exposure was calculated by the time spent at each postal code and successfully assigned to 65,254 study subjects with complete covariate information. 95  4.2.4.1 Air pollution exposure assessment Exposure to air pollution for each cohort member was assigned at their six-digit postal code(s), which corresponds to one block-face in urban areas, by three different approaches (described in detail in Brauer et al., 2008) namely, land use regression models, interpolation metrics of regulatory monitoring stations operated by the BC Ministry of Environment and Metro Vancouver, and proximity measures.  Land use regression models provided high spatial resolution (10 m precision), individual estimates of exposures to traffic-related air pollutants, including black carbon (BC), fine particulate matter (< 2.5 µm, PM2.5), nitrogen dioxide (NO2), and nitric oxide (NO) (Brauer et al. 2008; Henderson et al. 2007). Since the temporal stability of LUR surfaces showed reasonable backward predictions (Wang et al. 2013), each LUR model, developed after the 1999 – 2002 cohort inception period, was temporally adjusted based on daily air monitoring measurements to estimate monthly average concentrations that were assigned to individual subjects’ recorded home address postal code at birth (For detailed description, see (Brauer et al. 2008)) .  Data on NO2, NO, BC, PM10, carbon monoxide (CO) and sulfur dioxide (SO2)  from monitoring stations were used to compute an inverse-distance weighted (IDW) average of the three closest monitors within 50 km providing high temporal resolution (daily) estimates assigned to participants residential postal codes (Marshall et al. 2008b). Finally, road proximity for home postal codes of all cohort members was calculated using a geographic database (DMTI ArcView street file data set for BC, Canmap Streetfiles version 2006.3; DMTI Spatial Markham, Ontario, Canada). Home postal codes were categorized as within 50m or 150m of a primary highway, within 50 or 150m of a major road, or within 150m of a primary highway and within 50m of a major road. 96  4.2.4.2 Greenness Satellite-derived Normalized Difference Vegetation Index (NDVI) provided a continuous measure of greenness across the study region. NDVI ranges from -1 to 1 (with the minimum value indicating water and higher values indicating more greenness) based on land surface reflectance of visible (red: 0.4 – 0.7µm) and near-infrared (0.7 – 1.1 µm) parts of spectrum(Weier and Herring 2000). Landsat 7 Enhanced Thematic Mapper Plus (ETM+) cloud free images downloaded from 1999 to 2002 were used to assign greenness measures to study postal codes. Average yearly greenness values were calculated for 100 m areas around residential postal codes during pregnancy (Hystad et al. 2014). These buffer areas are at a similar spatial scale to within-city air pollution variation.  4.2.5 Analytical methods Nested case-control design using incidence density sampling within the study cohort was adopted for analytic efficiency. Each asthma case was randomly matched to five controls by sex and birth month and year. Nested conditional logistic regression models included breastfeeding status at the time of discharge, parity, maternal education (in area-level quartiles), household income (in area-level quintiles), gestational length and birth weight (both as continuous variables). The risk of incident asthma was examined in relation to air pollution and greenness in two time windows: between birth and up to the child’s 6th birthday (denoted preschool group thereafter) and from age 6 until the end of follow up (school age group). The preschool group mirrored the initial analysis of a subsample of this cohort (Clark et al. 2010b), while the school age group provided new insight into the effect of air pollutants and greenness on asthma incidence later in childhood. In addition, greenness and air pollutants were analyzed in joint exposure models. The effects of low birth weight (<2500g) and gestational duration (>37 weeks; 30 - 37 weeks; < 30 weeks) on asthma incidence were also examined in stratified analyses. Similarly, a series of stratified analysis were conducted to investigate potential effect modification by sex, parity, and socio-economic 97  factors (neighborhood-level household income < 40th percentile), maternal post-secondary education < median neighborhood level of maternal education, and tertiles of maternal age at birth).  4.3 Results A total of 6,948 and 1,711 children met the case definition of asthma onset in the preschool and school age groups respectively. Descriptive statistics by confounder data, stratified by case status are summarized in Table 4.1. Mean age at time of diagnosis (sd) was 2.64 (1.41) and 7.00 (1.21) years in the preschool and school age groups.             98  Table 4.1: Case and control characteristics (n (%) for categorical covariates; mean (standard deviation) for continuous covariates) for children meeting the case definition, matched controls by time window.  Preschool age School age  Control (34621) Cases (6948) Control (8577) Cases (1711) Sex                                   Female 13143 (38) 2646 (38) 3480 (41) 690 (40) Male 21478 (62) 4302 (62) 5097 (59) 1021 (60) Newborn Feeding at discharge     No breastfeeding 2430 (7) 545 (8) 626 (7) 141 (8) Breastfeeding  32028 (93) 6358 (92) 7911 (93) 1561 (92) Unknown 163 (0) 45 (0) 40 (0) 9 (1) Maternal post-secondary education quartiles   7670 (22)  1882 (27) 1959 (22) 459 (27) 1 (lowest) 2 7973 (23) 1622 (23) 1996 (23) 411 (23) 3 9407 (27) 1781 (26) 2270 (27) 421 (26) 4 9571 (28) 1663 (24) 2352 (28) 420 (24) Area-based household income quintiles   7468 (22) 1693 (24) 1833 (21) 392 (23) 1 (poorest) 2 7719 (22) 1697 (24) 1905 (22) 425 (25) 3 7178 (21) 1460 (21) 1893 (21) 377 (22) 4 6799 (20) 1185 (17) 1629 (19) 301 (18) 5 5457 (16) 913 (13) 1317 (16) 216 (13) Parity                                       0  15504 (45) 3332(52) 3879 (45) 892 (48) 1+ 19117 (55) 3616 (48) 4698 (55) 819 (52) Gestation length (weeks) 39.4 (1.15) 38.9 (1.94) 39.1 (1.68) 39.0 (1.8) Maternal age (years) 31.5 (5.06) 31.2 (5.09) 31.5 (5.12) 30.8 (5.3) Birth Weight (g) 3510 (458) 3362 (567) 3437 (527) 3385 (538) 99   Significant differences were observed for all risk factors in Table 4.1 between cases and controls for both the preschool and the school age groups. Asthmatic children were more likely to have been born to a younger mother, from a lower socio-economic stratum (as indicated by maternal educational attainment and household income), and less likely to have been breastfed. Also, the gestation period and birth weight were smaller among cases compared with their matched-controls.  All risk factors except for sex were similar when comparing matched controls from the entire pool of children who did not meet the case definition (i.e. all potential non-asthmatic controls), indicating non-biased selection of control subjects as shown in Table 4.2 Table 4.2 : Cohort characteristics (n (%) for categorical covariates; mean (standard deviation) for continuous covariates) for control children and all non-asthmatics by cohort age groups  Preschool group School group Controls (34621) No asthma (58306) Controls (8577) No asthma  (63543) Sex 13143(38) 31591(50) 3480(41) 30901(49) Female Male 21478(62) 33663(50) 5097(59) 32642(51) Newborn Feeding     No breastfeeding 2430(7) 4632(7) 626(7) 4491(7) Breastfeeding  32028(93) 60296(93) 7911(93) 58735(93) Unknown 163(0) 326(0) 40(0) 317(0) Female post-secondary education quartiles (1=lowest)  7670(22) 14441(23) 1959(22) 13982(22) 1 2 7973(23) 14776(23) 1996(23) 14365(23) 3 9407(27) 17792(26) 2270(27) 17371(27) 4 9571(28) 18245(27) 2352(28) 17825(28) 100   Preschool group School group Controls (34621) No asthma (58306) Controls (8577) No asthma  (63543) Area-based household income quintiles  (1=poorest)  7468(22) 13950(21) 1833(21) 13558(21) 1 2 7719(22) 14446(22) 1905(22) 14021(22) 3 7178(21) 17792(21) 1893(21) 13344(22) 4 6799(20) 12654(20) 1629(19) 12353(19) 5 5457(16) 10413(16) 1317(16) 10267(15) Parity 0 15504(45) 29602(45) 3879(45) 28710(45) 1 19117(55) 35652(55) 4698(55) 34833(55) Gestation (weeks) 39.4 (1.15) 39.1 (1.66) 39.1 (1.68) 39.1 (1.69) Maternal age (years) 31.5 (5.06) 31.4 (5.07) 31.5 (5.12) 31.4 (5.06) Birth Weight (g) 3510 (458) 3447 (533) 3437 (527) 3431 (533)   Table 4.3 summarizes the distribution of all spatially-derived exposure variables and the residential greenness metric for each nested case-control cohort. The levels of air pollutants in metropolitan Vancouver were relatively low, compared to other major Canadian cities and international urban areas with populations of over 1 million (Government of Canada 2007) , while the average greenness was high compared to other cities in Europe and the US with similar or larger size (Dzhambov et al. 2014) .          101     Table 4.3: Mean exposure levels, standard deviation (SD), and Interquartile Range (IQR) among cases and controls for carbon monoxide (CO), nitrogen monoxide (NO), nitrogen dioxide (NO2), ozone (O3), sulfur dioxide (SO2), particulate matter (PM10, PM2.5), and black carbon (BC) derived using inverse-distance weighting (IDW) and/or land-use regression (LUR) models and for greenness using Normalized Difference Vegetation Index (NDVI).   Preschool (0-5 years) School (> 6 years)  Pollutants_metric Mean IQR SD Mean IQR SD Cases CO_IDW (µg/m3)628.7 165.0 119.6 628.2 160.4 120.6 Controls 626.7 163.9 119.7 636.7 165.7 121.9 Cases NO2_IDW11 (µg/m3) 33.7 9.7 6.8 33.4 9.4 6.8 Controls 33.3 10.0 7.0 33.7 10.1 7.0 Cases NO_IDW  (µg/m3) 23.0 15.6 10.9 22.5 15.1 11.0 Controls 22.7 15.4 10.9 23.0 16.0 11.1 Cases O3_IDW10  (µg/m3) 27.7 8.6 5.8 27.9 8.7 6.0 Controls 27.9 8.5 5.8 28.3 8.6 5.9 Cases PM10_IDW  (µg/m3) 12.5 1.3 1.0 12.5 1.4 1.1 Controls 12.5 1.3 1.0 12.5 1.3 1.1 Cases SO2_IDW  (µg/m3) 5.7 3.2 2.4 5.5 2.8 2.4 Controls 5.6 3.1 2.4 5.7 3.3 2.5 Cases PM2.5_LUR  (µg/m3) 4.09 1.4 1.6 4.1 1.4 1.6 Controls 4.06 1.5 1.7 4.0 1.5 1.7 Cases NO2_LUR (µg/m3)  33.6 9.3 8.8 33.2 9.1 8.6 Controls 33.6 9.0 9.1 33.5 9.5 8.9 Cases NO_LUR  (µg/m3) 31.3 14.5 13.4 31.4 13.7 13.2 Controls 31.3 14.6 13.8 31.4 14.2 13.6 Cases BC_LUR  (10-5/m) 1.6 1.2 1.2 1.6 1.1 1.2 Controls 1.6 1.1 1.2 1.6 1.1 1.2 Cases Greenness_NDVI10 0.23 0.12 0.09 0.23 0.11 0.09 Controls 0.23 0.12 0.09 0.24 0.11 0.09 Air pollutants were positively correlated (except for ozone which was negatively correlated with other air pollutants) as reported previously (Clark et al. 2010b) and greenness was only moderately correlated with air pollution (r ≤ 0.4). The land use derived pollutants showed differences compared to the IDW derived pollutants both for children with incident asthma captured during preschool and school age periods, in particular for NO where the interpolation method was significantly lower (23 µg/m3 vs. 31 µg /m3 for                                                      11 Significant paired t-test comparison (p<0.05) 102  IDW and LUR respectively). The assigned exposure estimates were significantly lower among controls compared to their matched cases for NO2, Ozone, and greenness.  4.3.1 Asthma in preschool years For children with a diagnosis of asthma established in preschool years, all interpolation-based exposures were positively associated with an increased risk in adjusted models, except for ozone. No elevated odds ratios were found when using land use regression estimates (Figure 4.1A). Proximity to highway and major roads increased the risk of new asthma onset by 25% and 5% respectively (Table 4.4). Greenness at birth was protective as asthma risk decreased by 4% for each 0.11 NDVI increase in a 100m buffer around the residential home address at birth (Table 4.4). Consistent results for preschool children were seen for exposures categorized in quartiles (See Table 4.5 for exposure levels by quartiles). An increased risk was associated with increasing quartiles of exposure for all IDW exposures but ozone (Figure 4.1A), whereas no consistent trends were seen for quartiles of exposure from land use regression derived estimates. The associations with ozone were negatively correlated with asthma incidence due to the negative correlation of ozone with traffic-related air pollutants.         103  Table 4.4: Adjusted12 OR (aOR) for new asthma onset relative to an interquartile (IQR) increase in air pollutants (carbon monoxide (CO), nitrogen monoxide (NO), nitrogen dioxide (NO2), ozone (O3), sulfur dioxide (SO2), particulate matter (PM10, PM2.5)) derived using LUR and IDW, to road proximity, to surrounding greenness, and to joint greenness and air pollutants exposures.  Preschool  School Single Pollutants Unit IQR aOR (95%CI) IQR aOR (95%CI) NO_IDW µg/m3 15.5 1.06 (1.01 - 1.11) 15.5 0.93 (0.85 - 1.02) NO_LUR µg/m3 14.4 1.00 (0.97 - 1.03) 14.4 1.00 (0.94 - 1.06) NO2_IDW µg/m3 9.9 1.09 (1.04 - 1.13) 9.6 0.95 (0.88 - 1.03) NO2_LUR µg/m3 9.5 0.99 (0.96 - 1.02) 9.3 0.95 (0.90 - 1.01) CO_IDW µg/m3 163.9 1.05 (1.01 - 1.10) 164.6 0.90 (0.83 - 0.98) O3_IDW µg/m3 8.5 0.92 (0.87 - 0.97) 8.5 1.18 (1.07 - 1.31) PM10_IDW µg/m3 1.3 1.12 (1.05 - 1.19) 1.36 1.09 (0.96 - 1.24) PM2.5_LUR µg/m3 1.45 0.99 (0.97 - 1.01) 1.46 1.01 (0.97 - 1.06) Black carbon_LUR 10-5/m 1.13 1.01 (0.99 - 1.04) 1.13 0.99 (0.95 - 1.05) SO2_IDW µg/m3 3.1 1.05 (1.01 - 1.09) 3.1 0.89 (0.83 - 0.96) hwy5013 m  1.25 (1.04 - 1.49)  0.81 (0.55 - 1.19) mjr15012 m  1.03 (0.98 - 1.09)  1.04 (0.92 - 1.16) h150m5012 m  1.00 (0.94 - 1.08)  1.06 (0.92 - 1.21) Greenness_NDVI 0.11 0.96 (0.93 - 0.99) 0.1 0.99 (0.93 - 1.07) Joint Models  Preschool School NO_IDW µg/m3 15.5 1.05 (1.00 - 1.10) 15.5 0.92 (0.84 - 1.01) NO_LUR µg/m3 14.4 0.99 (0.95 - 1.02) 14.4 1.00 (0.93 - 1.07) NO2_IDW µg/m3 9.9 1.08 (1.03 - 1.13) 9.6 0.94 (0.87 - 1.03) NO2_LUR µg/m3 9.5 0.97 (0.94 - 1.01) 9.3 0.94 (0.88 - 1.00)                                                      12 Models were adjusted for newborn breastfeeding status at the time of discharge, parity, maternal education (in area-level quartiles), household income (in area-level quintiles), gestational length and birth weight. 13 Road metrics: hwy50: within 50m of highway; mjr150: within 150m of major road;  h150m50: within 150m of highway or 50m of major road 104   Preschool  School CO_IDW µg/m3 163.9 1.04 (0.99 - 1.09) 164.6 0.88 (0.8 - 0.96) O3_IDW µg/m3 8.5 0.93 (0.88 - 0.98) 8.5 1.22 (1.09 - 1.36) PM10_IDW µg/m3 1.3 0.98 (0.96 - 1.01) 1.36 1.09 (0.96 - 1.25) PM2.5_LUR µg/m3 1.45 0.98 (0.96 - 1.01) 1.46 1.01 (0.96 - 1.07) Black carbon_LUR 10-5/m 1.13 1.01 (0.98 - 1.03) 1.13 0.99 (0.94 - 1.05) SO2_IDW µg/m3 3.1 1.04 (1.00 - 1.08) 3.10 0.88 (0.81 - 0.95) hwy50* m  1.21 (1.01 - 1.45)  0.80 (0.54 - 1.19) mjr150* m  1.02 (0.96 - 1.08)  1.04 (0.92 - 1.17) h150m50* m  0.99 (0.93 - 1.07)  1.06 (0.92 - 1.22)  For both continuous and categorical exposures (see Table 4.5 for exposure levels by quartile range), all adjusted analyses were robust to the inclusion of the risk factors that were considered (breastfeeding, parity, maternal age, income quintile, maternal education level quartile, birth weight and gestational length) with odds ratio showing similar direction and magnitude of effects to that observed in the crude analyses.  Table 4.5: Exposure levels (N, mean) by quartile in the preschool age analysis among cases and controls   NO_IDW  NO2_IDW CO_IDW PM10_IDW SO2_IDW    Rank N mean N mean N mean N mean N mean Controls 0 8177 10.7 7254 23.7 7881 479.2 8439 11.1 7007 3.3 1 8444 16.6 9278 30.3 8904 570.3 9041 12.2 9291 4.2 2 9151 24.3 9164 35.4 8932 653.1 8730 12.8 9323 5.4 3 8845 38.1 8924 42.2 8894 787.3 8410 13.8 8999 9.2 Cases 0 1578 10.9 1302 23.8 1529 481.3 1680 11.1 1386 3.3 1 1702 16.6 1845 30.4 1835 569.9 1717 12.2 1843 4.1 2 1845 24.3 1913 35.3 1762 652.6 1807 12.8 1831 5.4 3 1821 38.1 1888 42.2 1821 788.8 1744 13.8 1888 9.1 105    NO_LUR NO2_LUR Black Carbon PM2.5_LUR     Rank N mean N mean N mean N mean  Controls 0 8157 18.2 7573 24.3 8397 0.7 7028 4.6  1 8689 24.9 8925 29.0 8739 0.9 7728 5.1  2 8910 31.6 9193 33.9 8916 1.4 7628 5.6  3 8865 49.3 8930 45.7 8569 3.5 7547 6.4  Cases 0 1591 18.3 1424 24.4 1654 0.7 1393 4.6  1 1791 24.9 1810 29.0 1743 0.9 1541 5.1  2 1773 31.6 1873 33.9 1761 1.4 1571 5.6  3 1793 49.1 1841 45.0 1790 3.5 1497 6.5     4.3.2 Asthma in school age years Children where incident asthma was only evident from age 6 onwards showed no consistent risk of developing asthma in relation to air pollutants exposures (Figure 4.1B). Most of the associations seen in early years were no longer evident, with the exception of exposure to PM10 that exhibited some consistency between the two time periods. The effect of ozone switched from having a protective effect during preschool to increasing the risk of developing asthma after 6 years of age (1.18 (1.06 – 1.31). No associations were found with regard to greenness exposure.  106   Figure 4.1: Adjusted Odds Ratio of incident asthma in children meeting the case definition in the (A) preschool and (B) school groups relative to exposure quartiles of  air pollutants (carbon monoxide (CO), nitrogen monoxide (NO), nitrogen dioxide (NO2), ozone (O3), sulfur dioxide (SO2), particulate matter (PM10, PM2.5))  derived by interpolation (IDW) and by land use regression (LUR) ; and  relative to measured greenness (NDVI).  4.3.3 Joint models  Given the moderately low correlation between air pollutants and surrounding greenness, joint exposure models of air pollutants with the greenness index were examined. Controlling for surrounding greenness only impacted the effect of PM10 on asthma risk in preschool children. The association between PM10 and asthma incidence disappeared from 1.12 (1.05 - 1.19) to 0.98 (0.96 - 1.01). For all other air pollutants and road proximity metrics, the direction and magnitude of effects on new onset asthma in young children as well as later in life did not change. On the other hand, the protective effect of surrounding greenness on asthma risk was increased after adjustment for air pollutants (Table 4.6).   107  Table 4.6: Effects of surrounding greenness on incident asthma risk, adjusted for air pollutants in preschool and school children.  Preschool 14 OR  (95% CI) School  OR  (95% CI) Greenness only  0.96 (0.93 - 0.99) 0.99 (0.93 – 1.07) + NO_IDW 0.83 (0.60 - 1.16) 0.77 (0.40 - 1.51) + NO_LUR 0.69 (0.49 - 0.97) 0.96 (0.48 - 1.91) + NO2_IDW 0.93 (0.67 - 1.3) 0.81 (0.41 - 1.59) + NO2_LUR 0.66 (0.47 - 0.92) 0.75 (0.38 - 1.48) + CO_IDW 0.82 (0.59 - 1.14) 0.68 (0.35 - 1.33) + O3_IDW 0.88 (0.63 - 1.23) 0.61 (0.31 - 1.19) + PM10_IDW 0.80 (0.59 - 1.1) 1.05 (0.56 - 1.97) + PM2.5_LUR 0.68 (0.49 - 0.94) 1.04 (0.54 - 2) + Black carbon_LUR 0.76 (0.55 - 1.05) 0.95 (0.5 - 1.8) + SO2_IDW 0.83 (0.59 - 1.15) 0.65 (0.33 - 1.27) + h150m50 0.70 (0.5 - 0.97) 1.08 (0.55 - 2.09) + hwy150 0.71 (0.51 - 0.98) 0.99 (0.51 - 1.91) + mjr150 0.72 (0.52 - 1) 1.07 (0.55 - 2.1)  4.3.4 Sensitivity analyses Stratified regression analyses were conducted only for the preschool period since this accounted for 70% of all identified cases and stratified models for the school age period did not converge. Among children who met the case definition during the preschool period, 535 (6.5% of all incident asthma cases) had between 30 and 37 weeks of gestational length and only 83 (1%) had less than 30                                                      14All models are adjusted for the following risk factors: maternal age at delivery, birth weight, gestational length, neighborhood-level household income, neighborhood-level maternal education, parity, newborn feeding at discharge 108  weeks, while 442 (6%) were below 2500 g at birth. Gestational length categories were grouped into a small gestational age category (<37 weeks) to ensure model convergence.  Suggestive evidence of effect modification by birth weight but not gestational age was demonstrated in stratified analyses(Figure 4.2). All the increased risks observed for children with birth weight less than 2500 g, for IDW-derived air pollutant exposures were near double that seen in children with birth weight more than 2500 g. Furthermore, stratifying by birth weight indicated consistent effects using land use regressions exposure estimates. NDVI in low birth weight children showed a consistent protective effect with near 50% reduction in the odds of new asthma onset.  Figure 4.2: Effect modification of greenness in 100m buffer and air pollution estimates derived using inverse distance weighting (IDW) and/or land use regression (LUR)models on new asthma onset among preschool age children by (A) gestational length (<37 weeks) and (B) birth weight (<2500g). All models are adjusted for maternal age at delivery, neighborhood-level household income, neighborhood-level maternal education, parity, and breastfeeding.  109  Stratified regressions by sex revealed that girls were consistently at increased odds of new asthma compared with boys for the same air pollutants identified in the main analysis and that they were more protected from the odds of new asthma onset with respect to the surrounding greenness compared to boys for whom the protective effect disappeared (Figure 4.4). Stratification by parity showed similar but less pronounced results: all exposures that had an effect on odds of asthma in the main analysis, except PM10, showed a heightened impact for children without siblings compared to those with siblings.  Figure 4.3: Risk of asthma incidence in the preschool group (n=6948), stratified by (A) sex and by (B) parity in relation to exposure to inverse distance weighting (IDW)- and land use regression (LUR)-derived air pollutants and surrounding greenness (NDVI) in 100m buffer. All models are adjusted for birth weight, gestation period, maternal age at delivery, neighborhood-level household income, neighborhood-level maternal education, parity, and breastfeeding  Lower socio-economic status (based on neighborhood income and education measures) children demonstrated high odds of asthma associated with interpolated air pollutants exposures (Figure 4.5). Stratification by maternal age at birth showed less consistent results: relative to interpolated air pollutants, both the oldest and youngest mothers were likely to have children at higher odds of new asthma development, while LUR estimates showed some trend with age 110   Figure 4.4: Stratified asthma risk analysis by three indicators of Socio-economic status (A: Household Income; B: Maternal Education status; C: Maternal age at birth) in relation to air pollutants and surrounding greenness in preschool age children . 111   4.4 Discussion  With 65,254 children followed up for 10 years, in one of the largest birth cohort studies to examine within-community air pollution contrasts, we found 25% increased odds of asthma for children living near highways and 5% to 10% increase for an interquartile increase in NOx and CO, pointing to traffic pollutant exposure in the etiology of asthma. Similar to an earlier analysis of a subset of this cohort (Clark et al. 2010b), we also found positive associations with PM10 and SO2 exposure (Clark et al. 2010b). The school age analysis revealed that exposures to air pollution generally did not affect asthma onset when the child was older than 6 years. Surrounding greenness elicited a robust protective effect in the preschool age analysis. This finding may follow a hypothesis that biodiversity loss( Rook 2013) and subsequent reduction in exposure to microbial diversity, reflected of less interaction with natural spaces, leads to immune dysfunction, inflammation and asthma (Haahtela et al. 2013). Greenness was not protective for asthma onset during school age years. This could be related to the mediating effect of physical activity. While this pathway is not likely to pertain to children between birth and age 5, it may come to play during school age years (Dadvand et al. 2014). Data on time activity patterns were not available, preventing exploration of such mechanisms. Correlations between air pollution and greenness levels were moderate, allowing the investigation of joint effects models on asthma incidence. During preschool years, the air pollution effects were largely unchanged after adjustment for greenness, whereas the protective 112  effect of greenness was strengthened after adjustment for air pollutants. This finding provides further support to the biodiversity hypothesis in that tolerance mechanisms are less impaired when interaction between human and environmental microbiotas are enhanced (Haahtela et al. 2013).  The increased risk of asthma incidence in early years was only evident with exposures estimated by interpolation methods (IDW). Although effect estimates based on exposures estimated by LUR models had smaller confidence intervals than those based on IDW, they showed no increased risk. There are several possible explanations for this discrepancy. First, these metrics capture different spatial scales (Marshall et al., 2008).  IDW may better represent secondary pollutants such as NO2 that display less spatial heterogeneity compared with primary pollutants (e.g. black carbon). Second, it is possible that LUR models, developed using 2003 measurements, may result in an over-smoothed surface for this cohort as exposure estimates were extrapolated backwards during a period time where concentration levels were decreasing Finally,  these differences between monitor-based and LUR model exposure estimates suggest some independence of risks between regional and local traffic-related air pollution (Brauer et al, 2008), however the positive associations observed with roadway proximity metrics would suggest otherwise.. The most recent birth cohort studies with similar follow-up periods show mixed results with regard to the role of air pollution. They include the Swedish cohort (BAMSE) where positive associations between air pollution and asthma after follow up until age 12 years were found (Gruzieva et al. 2013), whereas the British birth cohort found no association between NO2 nor PM10 and asthma in children who were 113  followed up until age 11 years (Mölter et al. 2014a). The number of identified asthma cases in BAMSE at increasing ages did not follow the decreasing pattern found in our study.    The Dutch Prevention and Incidence of Asthma and Mite Allergy (PIAMA) cohort study offers an interesting comparison (Gehring et al. 2010a). As in our study, positive associations between NO2 and asthma at age 2 and 4 (Brauer et al. 2007a; Gehring et al. 2002) were reported and most incident cases were apparent earlier in the children’s life. However, unlike our study,   the increased risk of incident asthma in relation to NO2 and PM2.5 exposure was still evident for incident asthma at older ages (up to age 8) in the Dutch study, possibly due to higher air pollution levels (Gehring et al. 2010a). In this investigation, the difference observed in the positive association between air pollution and development of asthma between birth up until age 6 and the absence of association later in life is likely to reflect that childhood asthma develops in early years rather than later in childhood. Additional analyses where the association was examined every year from birth to end of follow up confirmed that administrative records capture asthma incidence in the first 3 to 4 years of life leaving a very small pool of participants eligible for incidence analysis and thereby leading to a lack of power to detect any association between early life exposure and late childhood asthma development (data not shown).  Our asthma case definition, although verified in other population-based administrative data analyses (Kozyrskyj et al. 2004; To et al. 2006), relied on billing records for health care encounters and may have captured wheezing illnesses that have varying trajectories over time (Stein and Martinez 2004). Transient wheezing is common in infants and often resolves in childhood (Martinez et al. 1995a) as shown in a population-based Canadian study, where half the children diagnosed with asthma before age 6 went into remission by age 12 (To T et al. 2007). It is possible that the preschool analysis has included children whose symptoms may resolve later, weakening the ability to detect associations during school years. Future studies of asthma trajectories using linked health databases can help elucidate this question. 114  Administrative data are collected for non-research purposes and lack individual level information on important confounders (e.g., SES measure) and therefore may be subject to residual confounding. However, such linked health records provide unprecedented opportunities to investigate multiple risk factors in a large population. Preschool children weighing less than 2500g at birth were consistently, across all metrics, at increased risk for new asthma onset in relation to air pollution exposure, supporting the idea of greater susceptibility among low birth weight subpopulations (Strickland et al. 2014). However, the protective effect of surrounding greenness among this subgroup was 10-fold higher compared to children weighing more than 2500g at birth, raising the prospect of preventative public health interventions during pregnancy. Similar to the Children’s Health Study results (Salam et al. 2004), being born to a nulliparous mother decreased the risk of asthma incidence.  This observation may reflect the importance of the in utero environment, given evidence of reduced cord blood mononuclear cells proliferative responses with increasing parity (Devereux et al. 2002). We also confirmed previously observed effect modification of SES with air pollution's effects on childhood respiratory morbidity (Gehring et al. 2006). Further, older mothers were are at higher risk of  giving birth to children who were impacted by air pollution exposure, a relevant finding as British Columbia has the highest proportion of mothers giving birth over the age of 35 years in Canada (Canadian Institute for Health Information (CIHI) 2011).  In conclusion, air pollution was associated with incident cases of asthma in preschool children, but was not consistently related to initial asthma onset in school age children. For incident asthma during preschool years, surrounding greenness offered some protective effect independent from the observed associations with air pollution. 115   Chapter 5: Pediatric asthma patterns and effects of environmental factors.  5.1 Introduction Childhood asthma is a heterogeneous disease that encompasses a wide spectrum of presentations based on clinical characteristics or symptoms such as wheezing, especially in early childhood (Savenije et al. 2011). In order to conduct informative population-based studies, it is crucial to extend the current asthma cohort research beyond capturing incidence of a condition whose specific diagnosis remains elusive, to incorporate more detailed characterization of the different presentations of this disease.  The Tucson Children’s Respiratory Study (TCRS) was among the first studies to identify phenotypic groups for asthma based on clinical observations of wheezing (Martinez et al. 1995b): (1) non-wheezers, (2) transient wheezers (at least one wheezing symptom before age 3, no wheezing by age 6), (3) late onset wheezers (no wheeze until age 3, but wheezing by age 6) and, (4) persistent wheeze (wheezed in the first 3 years of life, and persistent wheezing at age 6). Subsequent studies have used latent class analysis to classify phenotypes of wheezing, complementing the initial categorization offered by the TCRS and contributing to clinical diagnosis of asthma (Henderson et al. 2008; Savenije et al. 2011; Spycher et al. 2008). The use of data-driven techniques - including but not limited to latent class modeling- to explore whether clusters exist within a population has been limited to a few studies of generally small sample size (Savenije et al. 2011) with the notable exception of one nationally representative birth cohort study of 11,652 British children (Panico et al. 2014), which relied on parental report of wheeze and atopic symptom (eczema and hayfever) episodes to classify four distinct trajectories. Since the use of self-reported conditions or symptoms is fraught with threats of bias, and given the logistical and financial burden of direct testing of children in large population-based studies, the use of administrative health records offers a unique opportunity to identify asthma trajectories, and to distinguish chronic and transient 116  conditions, thus capturing more specificity in relation to asthma phenotypes. Furthermore, with enhanced classification of asthma trajectories, we sought to conduct a novel evaluation of the role of environmental exposures on the different asthma groupings. In a population-based birth cohort study (the Border Air Quality Study (BAQS)) of over 65,000 children residing in Metropolitan Vancouver, British Columbia (BC), associations between air pollutants and asthma incidence were reported in children between birth and age 5, as well as an independent protective effect of surrounding residential green spaces, but not in the small proportion of cases whose disease developed between ages 6 and 10 (Chapter 4). It is plausible that the two age groupings that were examined represent different phenotypes, as identified in the TCRS groupings, with different risk. To further understand the role of air pollution and surrounding green spaces in childhood asthma etiology, the current study will examine annual asthma prevalence rates in the BAQS cohort to identify whether children follow different asthma trajectories that could not be captured in the prior analysis, and to understand how air pollution and greenness may affect the likelihood of belonging to the identified trajectories.  Group-based trajectory modeling is a specialized form of finite mixture modeling. The method is designed to identify groups of individuals following similarly developmental trajectories (Nagin 2005; Nagin and Odgers 2010). Unlike, hierarchical and latent class modeling, group-based trajectory modeling does not require parameters to be continuously distributed throughout the population according to the multivariate normal distribution. This approach assumes that there may be clusters of distinct developmental trajectories that may reflect unique etiologies. This offers the ability to model distinct subgroups of the population, allowing for the classification of children who belong to distinct trajectories with group-specific time-related parameters and facilitating the exploration of common risk factors and environmental exposures that may be associated with each asthma trajectory. 117  5.2 Methods 5.2.1 Cohort description Linked administrative data, accessible via the pan-provincial data services platform of Population Data BC (https:/www.popdata.bc.ca) were used to establish a birth cohort that comprised all single births from 1999 to 2002 to mothers with verified residential history in the metropolitan area of Vancouver, British Columbia, Canada  throughout their pregnancy (Brauer et al. 2008). This cohort included all residents with a primary address in BC as registration with the universal health insurance program is mandatory for those residing in the province for at least 3 months. Children born to these identified mothers were followed for 10 years (1999-2009) or until lost to follow-up, whichever occurred first. Loss to follow up occurred if they moved outside of study region, or when a gap in health registry exceeded 183 days. From the 73, 387 births to mothers with complete information on residential history during pregnancy, 988 multiple births were excluded, 101 children died during study follow up, and 4103 had missing covariate information leaving 68,195 singletons available for analysis. Population Data BC provides access to the Ministry of Health medical services  and hospitalization data  (British Columbia Ministry of Health 2009b) which inform the asthma definition, as well as vital statistics birth and deaths data (British Columbia (BC) Vital Statistics Agency 2009a, 2009b) which include birth date, sex, and First Nations status, obtained from the British Columbia Vital Statistics Agency. Population Data BC also links the preceding data with data from the BC Perinatal Data Registry (Perinatal Services BC 2009) which provides information on intention to breastfeed, maternal age, parity and maternal smoking during pregnancy. Since no individual-level data were available for socio-economic factors, census data on income and maternal level of education were assigned to children based on their neighbourhood of residence at birth at the Census Dissemination Areas (DA) level (Statistics Canada 2001), a unit representing a target population of 400 to 700 persons. The study protocol was approved by the Institutional Review Board of The University of British Columbia (#H04-80161).  118  5.2.2 Asthma case definition Asthma diagnoses were identified from physician billing and hospital discharge records (International Classification of Diseases [ICD]-9th revision: 493; ICD-10: J45) for the1999-2009 period. Using a validated case definition of asthma (Prosser et al. 2008), children with a minimum of two primary care physician diagnoses in a rolling 12-month period or a minimum of one hospital admission were identified as asthma cases each year from birth until end of follow up, providing an annual asthma prevalence.  5.2.3 Exposure assessment  Residential air pollution exposure during pregnancy was assessed using high resolution land use regression (LUR) air pollution models developed for the Vancouver metropolitan area for nitrogen dioxide (NO2), fine particulate matter (PM2.5), and black carbon (BC). The creation and evaluation of the models has been described previously (Brauer et al. 2008; Henderson et al. 2007; Larson et al. 2009) as has their application to the epidemiologic analysis of childhood respiratory disease (Clark et al. 2010b; Karr et al. 2009; MacIntyre et al. 2011, and Chapter 4) Measurements of NO2, NO, BC, PM2.5, PM10, carbon monoxide (CO) and sulfur dioxide (SO2) collected from regulatory air quality monitoring network stations were used to compute inverse-distance weighted (IDW) averages of the three closest monitors within 50 km. This provided daily estimates assigned to participants residential postal codes, then aggregated over the pregnancy period (Marshall et al. 2008b). A 6-digit residential postal code, in urban setting, refers to an urban  block face with a target population of 400 to 700 people.  The Normalized Difference Vegetation Index (NDVI) provided a continuous measure of greenness from -1 to 1 across the study region. Using Landsat 7 Enhanced Thematic Mapper Plus (ETM+) cloud free images downloaded from 1999 to 2002, NDVI derived from land surface reflectance of visible (red: 0.4 – 119  0.7µm) and near-infrared (0.7 – 1.1 µm) parts of spectrum (Weier and Herring 2000), was calculated for 100m buffers around residential postal codes during pregnancy (Hystad et al. 2014).  5.2.4 Analytical methods  The group-based approach utilizes a multinomial modeling strategy. Maximum likelihood is used for the estimation of the model parameters. The fundamental concept of interest is the distribution of asthma outcomes conditional on age (continuous) over time; that is, the distribution of outcome trajectories denoted by  where the random vector Yi represents individual i’s longitudinal sequence of outcomes (asthma diagnosis) and the vector Agei represents individual i’s age when each of those measurements is recorded. The group-based trajectory model assumes that the population distribution of trajectories arises from a finite mixture of unknown order J. The likelihood for each individual i, conditional on the number of groups J, may be written as   where pij is the probability of membership in group j, and the conditional distribution of Yi given membership in j is indexed by the unknown parameter vector βj that determines the shape of the group-specific trajectory. For a given j, conditional independence is assumed for the sequential realizations of the elements of Yi , yit, over the T periods of measurement, written as   where p(.) is the distribution of yit conditional on membership in group j and the age of individual  i at time t.  The binary logit distribution was used to to specify p(.) for the analysis of longitudinal data on a dichotomous outcome variable, here whether or not the subjects meets the asthma definition at age t.  120  In order to choose the appropriate number of groups, the model was repeatedly fit with a step-wise increase in the number of groups, starting with two groups fixing the order (i.e. shape) to quadratic for each group, since we assumed that there would be at least one non-asthmatic and one asthmatics group. As there is no single statistical test to determine the correct number of trajectory groups (Nagin 2005),  model fit was assessed using Bayesian Information Criteria (BIC) and log likelihood. A number of models with different orders in an interactive two-step process were tested: (1) selection of the number of trajectory groups, and (2) selection of the shape (also designated as order) for each trajectory group, using both statistical and subjective knowledge on asthma natural history. Following model selection, risk factors for the probability of membership in the asthma trajectory groups, relative to the non-asthma group, were tested using weighted multinomial logistic regression. Weights were calculated as products of trajectory group posterior probabilities. Combining trajectory groups with weighted regression methods can provide risk profiles for individuals most likely to follow certain trajectories over time. To build an overall prediction model, risk factors were selected in separate models using block testing. Block 1 included clinically important factors and known risk factors: sex, parity, First Nations status, birth weight, gestational age at birth, intention to breastfeed, household income, maternal education and smoking. Block 2 considered each of the environmental exposures separately (i.e. unadjusted exposure models). The final model consisted of the selected covariates from Block 1 and only significant environmental exposures from Block 2. 5.3 Results 5.3.1 Cohort and asthma trajectories description As the cohort inception included children born between 1999 and 2002, their contribution to years of follow up was examined. Participants were registered for nearly all of the study follow-up, with an average of 8.2 years (Table 5.1).  121  Descriptive statistics of the overall cohort (n=68195) with complete covariate and exposure assessment data (n= 65254) are summarized and stratified by trajectory group membership in Table 5.2. A total of 10,275 children (15 % of the sample) met the case definition during the course of the follow-up period. The prevalence of asthma cases rose monotonically until children reached age 5 up to 19.8%, at which point it steadily decreased until the end of the 10-year follow down to 5.2%. Table 5.1: Summary statistics of number of elapsed years of data coverage by birth year Cohort Birth year N Mean Min  Max 16979 9.58 0.00 10.00  Born in 1999  Born in 2000 17801 8.63 0.00 9.00  Born in 2001 17981 7.70 0.00 8.00  Born in 2002 15434 6.73 0.00 7.00  The majority of the cohort’s subjects were born after 37 weeks of pregnancy from non-smoking mothers (mean age at birth 31 years) who had initiated breastfeeding at time of discharge. Over half of participants (54.2%) were born to mothers with a previous live birth (Table 5.2). Birth weight at delivery was available as a continuous variable (mean = 3410g) but dichotomized as term birth weight less than 2500g for analysis, previously identified as an important modifier of the association between air pollution and asthma development (Sbihi et al, 2015 submitted). Similarly, gestational age at birth (mean: 39.1 weeks) was defined as binary variable using 37 weeks cut-off.     122  Table 5.2: Risk factor distribution (frequency counts for categorical variables; mean and standard deviation (sd)) by asthma trajectory groups among 68195 children.  Risk factors N (%) Non-asthmatics (n=63166) Transient asthma cases (n=2282) Late infancy onset chronic asthmatics  (n=1828) Early infancy onset chronic asthmatics    (n=919) Maternal Non-smoker  63597 (93.3%) 58843 (93.2%) 2139 (93.7%) 1739 (95.1%) 876 (95.3%) Native status   431 (0.63%) 402 (0.6%) 15 (0.7%) 8 (0.4%) 6 (0.7%) Female  33052 (48.5%) 31265 (49.5%) 779 (34.1%) 698 (38.2%) 310 (33.7%) Nulliparous   31309 (45.9%) 28919 (45.8%) 946 (41.5%) 994 (54.4%) 450 (49%) High SES   38335 (56.2%) 35778 (56.6) 1193 (52.3) 915 (50.1) 449 (48.9) High maternal education  37591 (55.1%) 35121 (55.6%) 1127 (49.4%) 903 (49.4%) 440 (47.9%) Newborn feeding at discharge  63015 (92.4%) 58443 (92.5%) 2055 (90.1%) 1691 (92.5%) 826 (89.9%) Birth weight ≥2500g  65696 (96.3%) 60947 (92.5%) 2160 (90.1%) 1742 (92.5%) 847 (89.9%) Gestation ≥37 weeks  64555 (94.3%) 59901 (96.5%) 2106 (94.7%) 1712 (95.3%) 836 (92.2%) Maternal age at birth (mean (sd)) 68195 31.3  (5.11) 31.2 (5.09) 31.0 (5.21) 31.1  (5.02)   The model with four different groups and cubic order was chosen as the best fit as it had the lowest BIC value. The four trajectory asthma groups that emerged from the group-based modelling (Figure 5.1) can be summarized as follow: - Non-asthma trajectory (NA):  88.8% of the study population has a quasi-null probability of meeting the case definition throughout the study follow up time. 123  - Late-Infancy Chronic Asthmatics (LI CA) trajectory: 4.1% of the study population have a non-remitting diagnosis throughout the 10 year follow up; in this group children start developing asthma around age 2, with a peak prevalence of asthma when they are 6 years old and maintain their asthmatic status up until end of follow up. This trajectory is akin to growing in to asthma phenotype.  - Early Infancy Chronic asthmatics (EI CA) trajectory: 1.5% of the population have a similar trajectory to that of the LICA trajectory as children maintain their asthmatic status until the end of follow up but after having met the asthma case definition in early infancy (before age 2) with a peak in prevalence around age 4. This trajectory is akin to continuing to have asthma symptoms. - Transient asthma trajectory:  5.6% of the study participants follow a transient pattern (i.e. grow out of asthma) where the asthma case definition is met in infancy with peak prevalence around age 2. No asthma activity is seen after children turn 6 years of age. 124   Figure 5.1: Group memberships and confidence intervals for the posterior distributions of probability of asthma at each time point between birth and age 10 for each of the trajectories in BAQS cohort. Trajectory 1 (in red) represents the late infancy onset chronic asthmatics (LI CA), trajectory 2 (in green) represents the non-asthmatics, trajectory 3 (in blue) represents the transient group, and trajectory 4 (in black) represents the early infancy onset chronic asthmatics (EI CA).    While only 919 children developed chronic asthma in early infancy, their trajectory was significantly different from that of the late infancy onset chronic asthmatics as indicated by the distinct confidence intervals around the predicted posterior probabilities (Table 5.3). Among children who met the case definition of asthma, transient asthmatics (n=2282) with resolved symptoms by the end of follow up represented the same proportion of the study population as those who were chronic asthmatics regardless of the period of asthma onset (5.6%).    125  Table 5.3: Posterior probability for each trajectory at each year of age (Non-Asthmtics; Transient; LICA: Late Infancy Chronic Asthma; EICA: Early Infancy Chronic Asthma) and 95% confidence intervals Year  p(Non asthma) p(Transient) p(LICA)  p(EICA)  0 0.01 [0.01 ; 0.01] 0.23 [0.21 ; 0.25] 0.00 [0 ; 0] 0.15 [0.13 ; 0.18] 1 0.01 [0.01 ; 0.01] 0.38 [0.36 ; 0.40] 0.00 [0 ; 0] 0.43 [0.39 ; 0.46] 2 0.01 [0.01 ; 0.01] 0.41 [0.39 ; 0.43] 0.03 [0.02 ; 0.04] 0.65 [0.62 ; 0.68] 3 0.01 [0.01 ; 0.01] 0.34 [0.31 ; 0.36] 0.14 [0.12 ; 0.16] 0.75 [0.73 ; 0.78] 4 0.01 [0.01 ; 0.01] 0.22 [0.21 ; 0.24] 0.31 [0.28 ; 0.33] 0.78 [0.76 ; 0.8] 5 0.01 [0.01 ; 0.01] 0.12 [0.10 ; 0.13] 0.41 [0.39 ; 0.44] 0.76 [0.73 ; 0.78] 6 0.01 [0 ; 0.01] 0.06 [0.05 ; 0.06] 0.42 [0.4 ; 0.45] 0.69 [0.66 ; 0.71] 7 0.00 [0 ; 0.01] 0.03 [0.02 ; 0.03] 0.37 [0.35 ; 0.39] 0.58 [0.55 ; 0.61] 8 0.00 [0 ; 0] 0.02 [0.01 ; 0.02] 0.29 [0.27 ; 0.31] 0.45 [0.42 ; 0.48] 9 0.00 [0 ; 0] 0.02 [0.01 ; 0.03] 0.22 [0.2 ; 0.24] 0.33 [0.29 ; 0.36] 10 0.00 [0 ; 0] 0.02 [0.01 ; 0.03] 0.18 [0.15 ; 0.21] 0.24 [0.19 ; 0.29]  All risk factors except for First Nations status were significantly different by group membership. Early infancy chronic asthmatic children were less likely to have been breastfed than non-asthmatics (10.1% vs. 7.5% respectively). Children in the non-asthma group were born to families with significantly higher neighborhood SES (household income and maternal educational attainment) compared to all other trajectories (Table 5.2). Both chronic asthma trajectories had more children born to nulliparous mother (54% and 49% for late and early infancy onset) than the transient cases and non-asthmatics (41 % and 46%, respectively). Following ANOVA, term birth weight and gestational period (when examined as continuous variables) showed significant differences across trajectories. These differences were driven by the contrast between non-asthmatics and children belonging to the three trajectories (i.e. Transient, Early Infancy Chronic Asthma (EICA), Late Infancy Chronic Asthma (LICA)). The former group has significantly larger weight and gestation length at birth. Further, within those meeting the case definition, the two chronic asthma group children had significantly smaller weight and gestation length at delivery compared to children belonging to the transient asthma group. Mothers of chronic asthmatics were significantly younger (mean (in years) (sd): 31.1 (5.1) and 31.0 (5.2) for early and late infancy onset respectively) than those of children who did not meet the asthma case definition (31.3 (5.1)). 126  The weighted multinomial regression relative to the risk factors available for analysis (Table 5.4) shows that smoking, sex, parity, household income, maternal education (high educational attainment being greater than median of neighborhood level), breastfeeding and birth weight had significant effects on the odds of membership in all of the three asthma trajectories.  Table 5.4: Odds Ratio (OR) of adhering to a trajectory group relative to risk factors based on weighted multinomial logistic regression15 Effect Category Group16  OR  (vs NA group) 95% Wald CI p-value Maternal smoking Yes vs No T 0.86 0.71; 1.03 0.10 Maternal smoking Yes vs No LI CA 0.66 0.53; 0.83 0.00 Maternal smoking Yes vs No EI CA 0.60 0.42; 0.84 0.00 Sex F vs M T 0.52 0.47; 0.57 <.0001 Sex F vs M LI CA 0.63 0.57; 0.70 <.0001 Sex F vs M EI CA 0.50 0.43; 0.58 <.0001 Parity 1 vs 0 T 1.21 1.10; 1.33 <.0001 Parity 1 vs 0 LI CA 0.70 0.63; 0.78 <.0001 Parity 1 vs 0 EI CA 0.89 0.77; 1.04 0.15 Maternal age 1st vs 3rd tertile T 1.14 1.02; 1.28 0.05 Maternal age 1st vs 3rd tertile LI CA 1.06 0.93; 1.20 0.11 Maternal age 1st vs 3rd tertile EI CA 1.15 0.96; 1.37 0.26 Maternal age 2nd vs 3rd tertile T 1.08 0.96; 1.20 0.91 Maternal age 2nd vs 3rd tertile LI CA 0.94 0.82; 1.06 0.09 Maternal age 2nd  vs 3rd tertiles EI CA 1.1 0.92; 1.32 0.71 SES H vs L T 0.91 0.83; 1.00 0.05 SES H vs L LI CA 0.82 0.74; 0.91 0.00                                                      15 χ2(30) <0.001, N= 64264 (3931 observations were deleted due to missing values for the explanatory variables)  16 T: Transient; LICA: Late Infancy Chronic Asthma; EICA: Early Infancy Chronic Asthma 127  Effect Category Group16  OR  (vs NA group) 95% Wald CI p-value SES H vs L EI CA 0.81 0.70; 0.94 0.01 Education H vs L T 0.81 0.74; 0.89 <.0001 Education H vs L LI CA 0.81 0.73; 0.90 0.00 Education  H vs L EI CA 0.77 0.67; 0.91 0.00 Breastfeeding No  vs Yes T 1.24 1.06; 1.44 0.00 Breastfeeding No  vs Yes LI CA 0.95 0.78; 1.16 0.64 Breastfeeding No  vs Yes EI CA 1.28 1.01; 1.63 0.04 Birth weight ≥2500 vs <2500 g T 0.75 0.59; 0.96 0.02 Birth weight ≥2500 vs <2500 g LI CA 0.79 0.59; 1.05 0.10 Birth weight ≥2500 vs <2500 g EI CA 0.49 0.35; 0.68 <.000 Gestation  ≥37 vs <37 weeks T 0.75 0.61; 0.92 0.01 Gestation  ≥37 vs <37 weeks LI CA 0.91 0.71; 1.17 0.48 Gestation  ≥37 vs <37 weeks EI CA 0.84 0.61; 1.16 0.28   5.3.2 Impact of environmental factors on asthma trajectories  Exposure metrics were successfully assigned to nearly all cohort members (Table 5.5) with complete covariate information (99.7%) except for IDW-derived ozone and fine particulate matter PM2.5 (95% and 85.6%, respectively).  The study area had low levels of air pollution with median NO2 of 33.3 µg/m3, a median inhalable particulate matter (PM10) concentration of 12.5 µg/m3, and a relatively high green space coverage (median NDVI =0.24).  128  Table 5.5: Exposure levels of greenness, IDW-derived pollutants and LUR-derived PM2.5, by trajectory among CHILD study participants Exposure Trajectory N Median Mean SD Min Max Greenness No Asthma 63017 0.24 0.23 0.09 -0.08 0.59 Transient 2271 0.23 0.23 0.09 -0.08 0.56 LI CA 1822 0.24 0.23 0.09 -0.05 0.56 EI CA 919 0.24 0.23 0.08 -0.04 0.49 Carbon Monoxide  No Asthma 62997 618.2 629.5 121.4 362.8 1062.3 Transient 2270 610.6 629.4 123.3 381 1021.6 LI CA 1820 613.3 626.5 117.8 384.8 1024.3 EI CA 919 609.0 621.2 113.8 395.9 983.2 Nitrogen Monoxide (µg/m3) No Asthma 63000 20.8 23.0 11.1 2.8 65.2 Transient 2271 20.4 23.0 11.3 2.9 63.2 LI CA 1821 20.3 22.5 10.6 3.6 61.0 EI CA 919 19.7 22.3 10.6 4.2 54.8 Nitrogen Dioxide (µg/m3) No Asthma 63012 33.3 33.5 7.0 15.0 53.7 Transient 2271 33.3 33.7 7.0 16.9 51.2 LI CA 1822 33.5 33.7 6.6 17.3 52.4 EI CA 919 33.1 33.4 6.5 16.9 51.0 Ozone No Asthma 60451 28.0 27.8 5.9 9.8 48.2 Transient 2162 28.1 27.9 6.1 10.9 45.2 LI CA 1758 28.3 27.9 6.0 11.4 46.7 EI CA 879 28.3 27.9 5.8 11.1 43.3 SO2 No Asthma 63012 4.7 5.7 2.5 1.9 17.8 Transient 2271 4.7 5.7 2.5 2.1 16.1 LI CA 1822 4.7 5.6 2.4 2.2 16.0 EI CA 919 4.7 5.5 2.3 2.2 13.8 PM2.5 No Asthma 54160 5.4 5.5 0.7 3.2 7.6 Transient 1952 5.4 5.5 0.7 3.8 7.2 LI CA 1507 5.4 5.5 0.7 3.5 7.2 EI CA 781 5.4 5.5 0.7 4.2 7.1 129  Exposure Trajectory N Median Mean SD Min Max PM2.517 No Asthma 63017 4.03 4.07 1.74 0 11.3 Transient 2271 4.06 4.07 1.58 0 10.5 LI CA 1822 4.05 4.11 1.59 0 10.5 EI CA 919 3.97 4.08 1.68 0 10.4  From all the exposures that were considered, only inverse-distance weighted NO2 and LUR-derived PM2.5 showed a significant impact on membership in one of the three asthma trajectories relative to the asymptomatic group; the effect of greenness was only seen for an interquartile increase in exposure. Table 5.6 presents the multinomial logistic unadjusted and adjusted regressions for the relative risk of being in a given trajectory compared to being in non-asthmatic trajectory for air pollution (NO2 and PM2.5) and greenness exposures. For a one quartile increase in exposure during pregnancy, NO2 significantly increased the risk of being in a chronic asthma group relative to non-asthmatics group by 51% and 25% for early and late infancy onset respectively (Adjusted Weighted Relative Risk Ratios (RRR) and 95% Confidence Interval (95% CI): 1.51 (1.22 – 1.89) and 1.25 (1.07 – 1.46) for early and late infancy onset). Relative to the non-asthmatic group, the transient asthma group was affected less by NO2 compared to those with chronic asthma trajectories (RRR (95% CI): 1.08(0.94 – 1.23)). The NO2 exposure gradient showed a decreasing effect in risk (from 51% to 23% for EICA and from 18% to 4% for LICA), yet it remained significantly positive for the early infancy chronic asthma group. In utero exposure to fine particulate matter was significant across all asthma phenotype trajectories as a one quartile increase in PM2.5 exposure resulted in a 28%, 21% and 22% increased risk of membership in the early and late-infancy chronic asthma and transient                                                      17 Since the interpolated fine particulate exposure was assigned to only 85% of the study population, we report the land-use regression estimates of PM2.5which covers 98% of the participants with complete covariate information and show the overall exposure for the two metrics  130  groups compared to the non-asthmatic group, respectively. The effect of PM2.5 was most prominent among the transient asthma group as the relative risk remained significantly associated after adjustment for other risk factors.  Residential greenness in a buffer of 100m surrounding the home location was associated with a decreased risk of belonging to a chronic asthma group for the highest exposure category only (RRR (95%CI): 0.88 (0.72 – 1.09) and 0.92(0.80 – 1.07) for early and late infancy chronic asthma, respectively). The likelihood for children to follow a transient asthma trajectory relative to non-asthma group was only observed for an interquartile increase in greenness exposure. Unlike PM2.5 and NO2 exposures which were robust to risk factors adjustment, the effect of greenness was less consistent (Table 5.6).   Table 5.6: Multinomial logistic regressions of unadjusted and adjusted single exposure models:  greenness (NDVI),nitrogen dioxide (NO2), and fine particulate matter (PM2.5) models by asthma trajectories in BAQS children born between 1999 and 2002.    Unadjusted Models  Adjusted Models Exposure Category Group18 RRR 95% CI RRR 95% CI NDVI 3 vs 0 T 0.88 0.88 1.00 0.91 0.79 1.04   LI CA 0.92 0.79 1.07 1.05 0.90 1.23   EI CA 0.88 0.72 1.09 0.99 0.80 1.22 2 vs 0 T 0.98 0.87 1.11 0.99 0.87 1.13   LI CA 1.2 1.04 1.38 1.28 1.11 1.48   EI CA 1.08 0.89 1.31 1.12 0.92 1.36 1 vs 0 T 1.0 0.88 1.13 0.98 0.86 1.12   LI CA 1.06 0.92 1.23 1.09 0.94 1.26   EI CA 0.96 0.78 1.18 0.94 0.77 1.15 NO2 3 vs 0 T 1.13 0.98 1.29 1.11 0.96 1.27   LI CA 1.11 0.95 1.3 1.04 0.88 1.22   EI CA 1.21 0.96 1.51 1.23 0.98 1.55 2 vs 0 T 1.08 0.94 1.23 1.02 0.89 1.18   LI CA 1.39 1.19 1.62 1.31 1.12 1.53   EI CA 1.27 1.02 1.59 1.31 1.04 1.64                                                      18 T: Transient; LICA: Late Infancy Chronic Asthma ; EICA: Early Infancy Chronic Asthma 131     Unadjusted Models  Adjusted Models Exposure Category Group18 RRR 95% CI RRR 95% CI 1 vs 0 T 1.13 0.98 1.29 1.1 0.96 1.26   LI CA 1.25 1.07 1.46 1.18 1.01 1.39   EI CA 1.51 1.22 1.89 1.51 1.21 1.88 PM2.5  3 vs 0 T 1.10 0.96 1.27 1.07 0.93 1.23   LI CA 1.03 0.88 1.21 0.95 0.81 1.12   EI CA 0.94 0.76 1.16 0.85 0.69 1.06 2 vs 0 T 1.32 1.15 1.50 1.24 1.09 1.42   LI CA 1.36 1.18 1.58 1.24 1.07 1.45   EI CA 1.11 0.90 1.37 0.99 0.80 1.22 1 vs 0 T 1.22 1.06 1.39 1.16 1.01 1.33   LI CA 1.21 1.04 1.40 1.13 0.97 1.32   EI CA 1.28 1.04 1.56 1.17 0.95 1.43  Except for parity, all covariates had similar direction of effect as that observed in the risk factors multinomial logistic regression with the largest magnitude observed among the early infancy chronic asthma group. These children were more likely to not have been breastfed, and to be born to a multiparous and non-smoking mother. Most notably, the relative risk to develop chronic asthma in early infancy more than doubled if the child’s term birth weight was less than 2500g (RRR (95%CI): 2.22 (1.69 – 2.90)). 5.3.3 Co-exposures models  Given the low correlation between greenness and air pollutants (Figure 5.2), we examined two co-exposures models where NDVI was entered with NO2 and PM2.5 respectively as well as NO2 with PM2.5  to examine the relative contribution of these environmental exposures.  132   Figure 5.2: Scatter plot matrix of air pollutants and greenness exposure by asthma trajectory and exposure distributions among children with one of the three asthma trajectories.  As shown in Table 5.7, PM2.5 and NO2 have independent effects on the likelihood to be in a chronic or transient asthma trajectory compared to the non-asthmatic trajectory. Since there were suggestions of protective effects of greenness in relation to the three trajectories, we also investigated the effect of greenness added to air pollutants in stratified analysis by NO2 and PM2.5 tertiles. Exposure to greenness was only borderline protective amongst the two chronic asthmatics groups in high levels of air pollution (2nd and 3rd tertiles of NO2 levels and 3rd tertile of PM2.5 levels; data not shown).     133  Table 5.7: Co-pollutants multinomial logistic regression of air pollution (IDW-derived NO2 and LUR-derived PM2.5), greenness (NDVI), controlling for  significant risk factors on the likelihood of belonging to transient , late infancy chronic asthma (LICA), or early infancy chronic asthma (EICA) in BAQS children. Trajectory Exposure  Category OR 95% CI EI CA NO2 1 vs 0 1.42 1.1 1.8 2 vs 0 1.20 1.0 1.5 3 vs 0 1.17 0.9 1.5 PM2.5 1 vs 0 1.14 0.9 1.4 2 vs 0 0.98 0.8 1.2 3 vs 0 0.86 0.7 1.1 LI CA NO2 1 vs 0 1.18 1.0 1.4 2 vs 0 1.29 1.1 1.5 3 vs 0 1.00 0.8 1.2 PM2.5 1 vs 0 1.12 1.0 1.3 2 vs 0 1.27 1.1 1.5 3 vs 0 0.97 0.8 1.1 Transient NO2 1 vs 0 1.10 1.0 1.3 2 vs 0 1.04 0.9 1.2 3 vs 0 1.10 1.0 1.3 PM2.5 1 vs 0 1.15 1.0 1.3 2 vs 0 1.22 1.1 1.4 3 vs 0 1.05 0.9 1.2  5.4 Discussion In this work, we report yearly prevalence of asthma in a population-based birth cohort study of 68,195 children followed from birth to age 10 in the Greater Vancouver metropolitan region. Three distinct trajectories of asthma were identified using Group Based modelling. Our model not only distinguished transient from chronic asthmatic children but also allowed for the differentiation within chronic asthmatic children of those who develop asthma in early infancy (EI CA) from those who develop asthma after age 3 years. While most of the previous literature has distinguished between transient and chronic phenotypes 134  of asthma such as wheezing measured via participants’ self-report, the present study offers new and objective insight into the development of asthma in early infancy and provides support, in a large population-based dataset,  to a similar classification (early versus late infancy asthma onset) reported in an American birth cohort of 689 children (Chen et al. 2012).  This work was motivated by an earlier investigation in the same population where associations between air pollution prenatal exposures and asthma incidence were found in preschool children (i.e. between birth and 6 years), but not for school age children (between age 6 until end of follow up) (Sbihi et al. submitted). One explanation for these previous findings was that most incident cases were captured in early years, leaving a small number of incident cases during the school age period. With the approach adopted here, we quantitatively confirmed this pattern. In particular, among the 2282 transient asthma cases identified in this analysis, only 2% of this group met the case definition after 6 years of age. Given the course of asthma development in the late onset asthmatic children (peak between 4-5 years of age and decline after age 6), the inability to detect associations between air pollution and asthma incidence during school age can be more confidently related to trajectories indicating that development of asthma occurs either in the first three years of age or resolves during school age years  Longitudinal studies have provided evidence for three wheeze phenotypes in childhood (Stein and Martinez 2004): (1) transient wheeze that is associated with lower respiratory tract illnesses in early life that may resolve by 6 years of age; (2) non-atopic wheezing occurring in the first 3 years of life that is associated with lower respiratory tract illnesses, which may persist beyond 6 years into chronic asthma, and is not associated with atopy; and (3) atopic wheezing that starts before 6 years, may persist into adulthood, is strongly associated with atopy, and has a more severe clinical course. We note a striking parallel with the three asthma trajectories unveiled in our objective analysis of administrative health data: the transient group mirrors the transient wheeze phenotype, indicating that there is a significant proportion of children identified through administrative health databases as asthma cases when they are 135  most likely transient wheezing children. The two chronic asthma trajectories also reflect similarities with the non-atopic and atopic wheezing groups. In the absence of doctor’s diagnosis, and without more detailed clinical evaluation or medication utilization as has been used in other cohort studies of the developmental trajectory of childhood asthma(Wu et al. 2012), we can only speculate that early infancy chronic asthma would capture non-atopic asthma while the late infancy chronic asthma would capture the atopic asthma subgroup.  Our study, which encompasses over 65,000 children followed over 10 years, is to the best of our knowledge, the first large population-based study to explore both the trajectories of asthma based on administrative health records and the impact of environmental predictors on different trajectories of childhood asthma. The data-driven approach of Group-Based Trajectory Modeling (GBTM) indicates that nearly all asthma-like phenotypes can be captured in early infancy. In addition, transient, early and late infancy onset of chronic asthma trajectories closely match those found in studies of asthma and wheezing using similar approaches of clustering individuals into distinct class such as latent class analysis (Chen et al. 2012; Granell et al. 2012). We note however that one study reported seven patterns for a similar follow up period by defining two-year asthma period prevalence(Wu et al. 2012), although the ad-hoc definition of period prevalence length which is neither data-driven nor supported by timing of biological mechanisms may be open to criticism.  Another particular strength and novel aspect of this study is in the methodology adopted following trajectory modeling. We have extended previous efforts to advance asthma cohort research through identification of classes of asthma based on individual patterns over time by quantifying the effect of air pollution, greenness and other risk factors on the probability of belonging to the identified classes. Despite its strengths, GBTM is not immune to the same model inference pitfalls (estimates error, validity) as generalized linear models in the context of an observational study (i.e. non randomized data). In order to reduce the possibilities for bias, we used posterior probability as weights in the regression analyses 136  when comparing the “non-disease” to the diseased (Nagin and Odgers 2010). Combining trajectory groups with weighted regression methods provided risk profiles where elevated exposures to NO2 and PM2.5, two markers of traffic-related air pollution, were shown to independently increase the probability of adhering to a chronic pattern of disease, especially for infants (EICA group). Another important gap that was informed relates to the modest propensity of early life green spaces exposure to mitigate the increased risk for infants to grow into a chronic asthma trajectory following high levels of exposure to traffic-related air pollutants such as NO2. Despite the fact that children may grow out of their asthma-like symptoms, this work identified that greenness may provide some preventive measure given the decreased relative risks that were found.  Recent advances in group based trajectory modeling include the ability to follow the developmental course of two distinct but related outcomes. Future studies can benefit from such a dual trajectory model as it provides a rich statistical summary of the developmental linkages between the two outcomes of interest such as asthma and other comorbidities that evolve contemporaneously (e.g. bronchiolitis and asthma) or that evolve over different time periods that may or may not overlap (e.g., asthma during childhood and school achievement in school age year or during adolescence). In conclusion, trajectory modeling of physician’s diagnosed asthma during a 10 year since birth identified three distinct groups with direct assessment of disease status, that closely resemble asthma phenotypes identified in previous studies. This investigation indicated that an equal proportion of children diagnosed at least once during follow-up will not remit and develop chronic asthma; this condition (i.e. chronic asthma, whether onset is in early or late infancy) which is affected by both intrinsic and behavioral risk factors is significantly associated with early life traffic-related air pollution, especially among the group who develop chronic asthma in early infancy.   137   Chapter 6: Discussion and conclusions  The work presented in this thesis addresses methodological questions for researchers investigating the relationship between early life environmental exposures and the development of asthma and allergies in childhood. These questions are mostly centered on exposure assessment challenges. I investigated a novel measurement option and modeling methods where I adopted pragmatic approaches for exposure assessment in the context of individual-level temporally varying exposure estimates in large cohort populations. The contribution of other neighborhood exposures that may co-vary with air pollution was also examined by the inclusion of surrounding green spaces, and its impact on asthma was quantified. The different contributions summarized here inform questions such as: “What should we measure in air pollution health studies to maximize the precision of exposure estimates?,” “How can information gathered in the context of observational birth cohort studies help the estimation of personal exposures?,” and “What other co-varying environmental exposures may mitigate the adverse effects of air pollution on pediatric respiratory health?”.  A subsequent body of work in this dissertation focused on the complexity of the asthma syndrome that potentially covers multiple discrete entities of the disease. This work concluded by demonstrating the need to advance epidemiologic cohort studies on pediatric asthma and allergies by the examination of incidence of these heterogeneous chronic conditions to the investigation of their patterns and the role of risk factors over time.  138  6.1 Summary and contributions 6.1.1 Key findings The overall main conclusions of this dissertation are: (1) Early lifel exposure to traffic-related air pollution increases the risk of childhood asthma and atopy in the first years of life. The finding for atopy is novel in its analysis of atopy development in early infancy. The finding for asthma development is in line with a recent pooled analysis of several European and Canadian birth cohorts, the Traffic, Asthma and Genetics Study (TAG), where effects of TRAP were associated with the development of childhood asthma (MacIntyre et al. 2014) as well as with a meta-analysis that demonstrated a significant increased risk for asthma incidence (Anderson et al. 2013a). (2) The risk for asthma initiation following early life TRAP exposure is diluted over time, most likely due to the heterogeneity of the disease. However, TRAP exposure increases the odds of having an early-life onset chronic asthma course.  (3) Assessing traffic-related air pollution (TRAP) for large birth cohort studies based on modeled surrogate measures of TRAP such as NO2  that are spatio-temporally adjusted is warranted, while the use of hopanes in house dust was found to be of limited utlity.  (4) Surrounding residential greenness may offset the adverse effects of traffic-related air pollution on asthma incidence in the early years of life. In the following subsections, the contributions related to the above conclusions are outlined. 6.1.1.1 Are hopanes a useful metric for infiltrated traffic-related air pollution?  As detailed in the background review of the literature (Chapter 1), there is a need to identify air pollution exposure metrics that can account for infiltration of air pollutants and integrate over longer time periods than are feasible for air sampling. Hopanes, which have been shown to be good tracers of primary vehicle 139  exhaust, can be detected in indoor house dust, suggesting possible use as a surrogate for personal exposure to infiltrated traffic-related air pollution. After comparing measured dust hopanes with ambient air hopanes measurements and with GIS-derived land use variables for each city, the signature profile of ambient hopanes was significantly correlated, albeit modestly, with that found in home-settled dust. A key feature of this particular finding was its external validity, considering that the data, analyzed rigorously using the same protocol in all cities, was gathered in a varied setting ranging from highly urban locations (e.g., Toronto) to smaller cities (e.g., Winnipeg). Indoor measurements of hopanes levels in house dust were not correlated with GIS-derived land use variables unless behavioral and/or home characteristics were considered. This lack of correlation was evident whether the GIS-derived variables were generic land use predictors or variables more targeted to the land use specificities of each city. Examining predictors of LUR models rather than predicted NO2 estimates of LUR is a more precise approach to assess the utility of hopanes in settled dust as indicators of traffic-related air pollution exposure because NO2 estimates have a certain amount of error associated with the proportion of variability unexplained by the LUR models.  Taken together, the use of hopanes as a potential advance over existing traffic-related air pollution exposure assessment approaches was found to be limited, as hopanes only appeared to be useful if the determinants of house-settled dust variability were also characterized. These findings led to examining other approaches, such as modeling (as performed in the subsequent studies of this dissertation) to account for the aspects modifying total personal exposure (indoor/outdoor infiltration, time-activity patterns) that hopanes would have captured if this initial investigation had proved more successful. The subsequent sections highlight key findings when using modeling approaches of personal exposures.  140  6.1.1.2  Sensitization in infants and traffic-related air pollution  The work undertaken in the context of the CHILD study illustrated the utility of incorporating information from different sources to improve exposure assessment as a means to increase understanding of vulnerable time windows of exposure during which air pollution may impact the development and maturation of the immune system, and thus affect the onset of childhood atopy and, possibly, asthma. Typically, LUR models are used to assess individual exposures, and studies often rely on the residential birth address for their exposure assessments. Aspects such as the infiltration of outdoor air pollution indoors, residential mobility, and personal time-activity patterns may be important modifiers of personal exposures. Approaches to account for these factors, as well as temporal trends in outdoor pollution concentrations, were developed to estimate TRAP exposures (Chapter 3). These estimates were then applied in the analysis of association with sensitization among 2482 infants who were at least one year of age. Questionnaires, home inspections, governmental air-monitoring measurements, and geographic predictors were used to: account for temporal changes and residential mobility, examine indoor infiltration of TRAP based on a season-specific model, and facilitate stratified analyses based on time activity patterns. TRAP exposure during the first year of life but not during pregnancy increased the risk for development of atopy for any of the ten allergens that were administered during skin prick tests. The associations were largely driven by sensitization to inhalant allergens rather than food allergens. When residential mobility was not considered, a weaker association was found (aOR = 1.10; 95% CI: (0.96 – 1.34) first year) compared to the odds of developing atopy after adjustment for all addresses where participants resided (1.16 (1.00 – 1.41)). This association was stronger among those for whom the exposure was less susceptible to misclassification, i.e., participants spending more time in the home and participants not attending daycare. However, there may be an alternative explanation other than misclassification. A portion of the difference found in the sensitivity analyses of time-location patterns was likely attributable 141  to exposures other than TRAP. For example, analysis stratified by presence of siblings indicated that those with older siblings in the household were less likely to develop an allergic reaction to the skin prick tests. This explanation was also likely to hold for the analysis stratified by daycare attendance. Our results regarding indoor PM infiltration were inconclusive, showing an increased risk in homes with higher ventilation only during periods where the weekly average temperature was below 18°C. In an effort to explain the absence of effect modification in the warm season, attention was given to the possibility of an infiltration threshold being surpassed. All the homes, except those in Toronto, experienced significantly greater Finf in the warm compared to the cold period. This could imply that all homes (even those below the city-median infiltration efficiency) are in fact still “leaky”. In addition, as the challenge posed by infiltration mostly pertains to its variability over time, the variability in predicted Finf during the warm season was found to be much smaller in all CHILD cities, introducing misclassification error that would not necessarily affect health estimates, but merely modify the outdoor concentrations by a constant. These additional investigations and possible explanations argue in favor of developing Canadian-specific infiltration models. In this work, temporal adjustment of land use regression estimates at all residential home locations where the mother and child lived for at least 15 days captured associations that otherwise would have remained veiled. Through the use of a more refined methodology, we added new evidence to the existing TRAP and allergies literature. This conclusion is in line with the findings seen in a Dutch birth cohort study, where positive associations were demonstrated among a subset of non-movers (Gehring et al. 2010b) , suggesting that reducing exposure misclassification can help clarify the adverse role of TRAP on allergic diseases. The trajectory of atopy is likely to be heterogeneous (Simpson et al. 2010): longer-term studies can help clarify the role of TRAP in the etiology of allergies. With this perspective, the next step of this dissertation took advantage of a 10-year follow-up in the BAQS birth cohort to investigate the effects of TRAP on the initiation of childhood asthma, as atopy is a known risk factor for IgE-mediated asthma.  142  6.1.1.3 Childhood asthma and traffic-related air pollution The follow up over 10 years in BAQS, a population-based birth cohort, extended the initial investigation on pediatric asthma in early years (Clark et al. 2010b) by examining the effects of TRAP at different ages using the LUR modeling approach, among other exposure metrics, and adjusting for temporal and spatial variability (Chapter 3). Starting with the premise that early life exposure is a critical window, issues related to the time-course of developing chronic conditions such as asthma were evaluated. In addition, the role of potentially co-varying environmental exposures (air pollution and surrounding greenness) in asthma initiation was examined.  The investigation of incident asthma showed that air pollution was associated with the odds of asthma in children during the ages between birth and age 5, but not at older ages. These associations with air pollution during the preschool period were partially offset by the independent protective effect of the level of residential greenness surrounding the home location (Chapter 4). In an effort to better understand the course of asthma defined using administrative databases, annual incidence rates were also examined, and confirmed that most new asthma cases occurred in the first three years of life. To further evaluate the time course of childhood asthma in the BAQS cohort and to assess whether effects were focused on specific time periods, trajectories of asthma phenotypes were determined. Chapter 5 described two trajectories of children with chronic asthma: those who developed the disease starting in early infancy and those who started to show healthcare utilization for asthma treatment and/or diagnosis around age 5 or 6, coinciding with the cut-off year between preschool and school-age time periods that were examined in Chapter 4. The effect of TRAP exposure was most prominent among the chronic asthmatics who developed the disease in early infancy.  143  6.1.2 Synthesis The results in Chapters 2 and 3 of this dissertation provide different perspectives on the complexity of assessing personal air pollution exposures that occur mostly indoors, while the vast majority of the assessment tools at our disposal estimate ambient concentrations. These two chapters provided a measurement and a modeling approach to address the indoor infiltration of air pollutants.  Assessing indoor levels of TRAP through the collection and analysis of hopanes in settled house dust is a new area of study that has the potential to reduce exposure misclassification and increase specificity of personal exposure by capturing the infiltrated portion of the pollutants related to asthma and allergies. While the utility of hopanes as markers for infiltrated TRAP exposure were inconclusive, Chapter 3 tackled the issue of infiltration by examining whether modeled infiltration modified the association between air pollution and sensitization. This analysis indicated that modeled infiltration was partially useful given the observed modifying effect on the association between TRAP exposure and sensitization only during the winter season when ambient temperature was below 18ºC. However, several limitations should be described here. First, homes were classified according to the city-level 80th percentile of individual homes’ particle infiltration efficiency, aggregating individual-level data and thus introducing misclassification error that affects the precision of the estimates. Second, infiltration efficiency was determined using a model calibrated among homes included in an air pollution cohort study from the USA (Allen et al. 2012b). While this model explained 60% of the variance in infiltration efficiency during the summer and winter, the unexplained variability is likely to propagate error in our investigation and, more importantly, it was not validated for Canadian homes, possibly introducing measurement error. The effect of infiltration remains an open question because we have shown that the contribution of ambient TRAP to the indoor mix is likely to modify the total exposure.  The stratified analyses accounting for locations other than the homes, which indicated potential exposure misclassification, corroborated the importance of developing methods for estimating exposure in all 144  microenvironments (Chapter 3). The likelihood of misclassification when other microenvironments are not considered is supported by results shown in the cohort study in the UK (Mölter et al. 2013). In this study, investigators developed a model to integrate home, school, and journey exposures to assess children’s total individual exposure (Mölter et al. 2012a) for an epidemiological analysis of lung function among school age children (Mölter et al. 2013) and demonstrated a high correlation with personal measurements. The degree of misclassification in the presented sensitization analysis was probably non-negligible because infants in their first year of life tend to spend the vast majority of time in their homes, but accounting for residential histories and temporal trends provided reasonable personal exposure estimates. Moving from a smaller birth cohort with detailed information to a large population-based birth cohort without the same level of detailed individual information, investigations of pediatric asthma outcomes were conducted using temporally adjusted ambient air pollution levels at the residential location from conception to birth (Chapters 4 and 5). Chapter 3 indicated that early life exposures are critical (Clark et al. 2010a) as it demonstrated that with increased temporal and spatial specificity, the impact of exposures within narrower time windows (in utero vs. first year post-partum) can be pinpointed and their relative contribution in the etiology of allergic sensitizations uncovered.  Although the same study design was adopted in most of the work undertaken (Chapters 3-5), the birth cohort studies that were examined had varying degrees of information regarding potential effect modifiers on the health effects of air pollution. In the BAQS study, a notable finding was the effect modification by term birth weight. The identification of this susceptible subgroup (children weighing less than 2500g at term birth) is of clinical importance and is in line with previous studies reporting the susceptibility of premature children to pediatric lung disease development (Jobe and Tibboel 2014; Strickland et al. 2014). However, in the BAQS study several individual-level confounders were not accessible, limiting the ability to examine in detail other potentially susceptible populations. For instance, children of asthmatic 145  or atopic parents are more susceptible to TRAP effects on the development of allergic diseases (Morgenstern et al. 2008). Maternal atopic status was a major predictor in the atopy analysis; a confirmation of this finding in a population-based study would have given more weight to this identification of a susceptible group. In the same vein, the BAQS study may have been limited in its ability to investigate how socio-economic variables may affect the impact of air pollution on asthma development in children. Although household income and educational attainment were considered at the neighborhood level, residual confounding still remains. However, recent evidence has shown that area-level SES should also be considered in studies of air pollution and allergies (Gruzieva et al. 2014), which alleviates this threat of confounding.   In contrast, the CHILD study provided very detailed individual information. In particular, the physical characterization data allowed the examination of an identified gap in the literature on assessing personal exposure to air pollution with regard to the development of childhood asthma and allergies. Although the stratification by home ventilation level is novel and may help advance the field of air pollution exposure assessment, the process of stratification reduced the sample size, which, in turn, reduces the ability to detect small-magnitude effects. This limitation also applies to the stratification by daycare attendance and time-activity patterns. In these latter analyses, individual time spent away from the house was averaged across participants by city to stratify the cohort into two groups, introducing a Berkson-like type of misclassification.   Berkson- and classical-type errors constitute the two general classes of measurement error that are unavoidable in exposure assessment. Berkson errors generally introduce no bias in linear – and, to some extent, generalized linear – health effect models because they measure part of the “true” exposure, whereas classical errors measure the true exposure plus varying degrees of noise, thereby introducing some bias away or toward the null (Sheppard et al. 2011). In air pollution cohort studies, the impact of 146  measurement error is an active area of research because these two types of error come into play: the use of prediction models to estimate exposure will likely reduce variability (smoothing effect, “Berkson-like” error) and the spatial correlation of the exposure model parameter estimates may bias the health effect parameter estimation (spatial misalignment, “classical-like” error) (Dominici and Cefalu 2015).  In addition, the analysis of daycare attendance could have been strengthened by geocoding the addresses of daycares or similar sites where each child spent time outside the home, assigning exposures at these addresses and creating an average total exposure profile. At the time of data acquisition and cleaning, however, the number of missing values hindered this approach, leaving stratification as the next-best way to examine effect modification by exposure at locations other than the home.  Given the lack of specific markers, most of the birth cohort studies of pediatric respiratory outcomes examine inhalable particles, fine particles, or nitrogen dioxide. This dissertation has mostly focused on nitrogen dioxide, for which recent efforts to establish a dose-response formulation were developed based on the prolific set of population health studies in which NO2 is the principal exposure of interest (Favarato et al. 2014). Our results show that variation in NO2 exposure consistently predicts childhood asthma and sensitization, a finding that is aligned with the view that morbidity associated with ambient nitrogen dioxide is not only explained by its associated toxicants, but also with NO2 itself (World Health Organization Regional Office for Europe 2013). Although Chapters 3 and 4 examine the incidence of two different health outcomes, it is important to keep in mind that: (a) sensitization and asthma can be interrelated and (b) the two studies both used land use regression modeling of NO2 concentrations, a common exposure. These two chapters are complementary in the sense that one examines an unselected population-based birth cohort, the BAQS cohort, while the second utilizes primary data gathered in a non-random birth cohort of participants recruited by the CHILD researchers. In CHILD, considerable efforts were deployed to gather a large amount of data 147  through questionnaires, biological samples, and home inspections. The inclusion of such data is expected to enhance the precision of individual exposures estimates, despite the potential introduction of additional sources of error. This level of in-depth individual information is not accessible in study designs that rely on linked administrative databases. On the other hand, the BAQS studies (Chapters 4 and 5) offered the opportunity to investigate childhood asthma in a large sample of over 65,000 children, with objectively assessed exposures and diagnoses of asthma based on health care utilization records. Administrative databases such as these offer great promise in epidemiological research because they provide a comprehensive, longitudinal, and population-based record of health information that eliminates issues of self-report, recall, and selection bias inherent in survey-based research.  Collectively, this work demonstrated the importance of early-infancy exposures, a critical time window, and the adverse effects of TRAP on asthma and sensitization incidence, as well as the potential health benefits provided by surrounding green spaces. The associations demonstrated in the presented studies point to the heterogeneity of the asthma syndrome given the dissolution of effects over time, as well as to the complexity of atopy status definition. This heterogeneity begs for a different research framework, such as finite mixture models, to advance our understanding of the role of environmental exposures in the etiology of these chronic respiratory conditions. Our results also confirm the recently reported increased susceptibility of children with parents reporting an allergic disease (Gruzieva et al. 2012; Morgenstern et al. 2008).  6.1.3  Strengths and limitations Birth cohorts offer multiple advantages in the assessment of chronic disease initiation in relation to environmental exposures. In both BAQS and CHILD, participants were prospectively followed from conception onward, providing the best setting for the investigation of the temporal sequence of events and 148  indicating the direction of causality. The present dissertation does not prove causality, but it adds to the weight of evidence for a causal relationship between TRAP and incident asthma and allergies in young children. An additional strength of this body of work is the complementarity between the study populations that were leveraged to examine the effect of traffic-related air pollution on asthma and allergies. Each of the two cohort studies, which included outcomes that share some common etiology, provides aspects that are lacking in the other (e.g. size, individual-level characteristics). As a whole, they strengthen the weight of the reported findings.  BAQS is a population-based cohort, which, in addition to its large sample size, offers a sample with good external validity, whereas CHILD may have children with parents of a higher SES than the overall Canadian population. It is therefore unclear to what extent the results reported from CHILD study are generalizable to a low-SES segment of the Canadian population. In addition, other than moving out of the province, the BAQS studies do not suffer from the same threat of selection bias attributable to loss of follow-up that is inherent to a birth cohort study such as CHILD.  6.1.3.1 Assessment of environmental neighborhood exposures  Environmental exposures linked to the development of childhood allergies and asthma occur from conception through the prenatal period and into childhood, with the latest evidence pointing at either the first or second year of life as the critical time window (Arrandale et al. 2010) due to the immaturity of the lungs and immune system. The question of when to measure these exposures is complex, and consequently one of this dissertation’s main strengths resides in the utility of the methods that were employed to assign exposure for precise time windows.  149  Air pollution exposure assessment in the present dissertation was estimated for the early life period using previously developed and validated LUR models. These types of models have been widely used throughout the past two decades of air pollution epidemiological studies without assessing whether they are valid for study periods that are distinct from the time when the models were developed, thereby incorrectly assuming that spatial patterns are stable. In the context of this work, NO2 concentrations have decreased over time in all of the cities examined in this dissertation (Brook et al. 2014). In Vancouver, it has been demonstrated that when NO2 concentrations decrease over time the spatial concentration patterns remain relatively stable (Wang et al. 2013), and thus temporal adjustments using government fixed-site monitoring are appropriate.  LUR models in the present work are a strong tool for generating individual exposure estimates. They provide a high spatial resolution that was augmented by tracking the residential histories throughout pregnancy and during the first year of life, in comparison to previous studies that solely relied on estimates of traffic-related air pollution at the home address at time of birth. Accounting for residential mobility over a time period of increased susceptibility helps attenuate the misclassification that would have occurred if exposure was estimated solely using the birth address. While exposure levels around the early life  period were of primary interest, this work would have benefited from examining the relative importance of exposure effects at birth address versus current home address, and confirming earlier findings such as those shown in the BAMSE birth cohort studies (Gruzieva et al. 2012, 2013), which indicate that the early years are a critical time window for asthma and allergies.  Another important strength of the present work is the examination of the effect of exposures that occur away from the home. The inclusion of all of the places in which a young child spends time during the first year of life should reduce exposure misclassification. Among school-age children, McConnell and colleagues showed that TRAP exposure at school and home may contribute to the development of asthma (McConnell et al. 2010), whereas in younger children, TRAP exposures in different locations have been 150  considered rarely, so the present work fills an important gap. Although exposures were not quantitatively assigned, we showed that time-activity patterns are likely to modify the association between TRAP and the development of allergies in early-life. An additional strength of the environmental exposure assessments is the adjustment of air pollution for the presence of green spaces in the BAQS cohort (Chapters 4 and 5). These two exposures were only moderately correlated, enabling the examination of co-exposure models and the demonstration that air pollution affects respiratory outcomes independently from the effect of green spaces. Therefore, the assessment of multiple exposures related to urban form (various air pollutants and NDVI-based greenness) is an important contribution of this work. While the Normalized Difference Vegetation Index (NDVI) offers an objective assessment of green spaces, it remains a non-specific measure of the type of greenness surrounding the homes of study participants. Dadvand et al. showed that different types of greenness have differential risk and health benefits (Dadvand et al. 2014). The mechanisms of access to green spaces on pediatric respiratory health can be explained by the biodiversity loss hypothesis, increased aero-allergic load, or the reduction in air pollution. Small buffer sizes were used around the home address, decreasing the likelihood of varying degrees of allergenicity of the surrounding greenness. However, the last set of analyses involving greenness at birth, even when measured using NDVI in small buffer areas rather than relying on an estimate of plant species by differing degree of allergenicity, is not a relevant metric for effects later in childhood. Other environmental exposures that may be relevant to this work but were not considered include the effect of climate (e.g., variations in precipitation, temperature, and relative humidity). A major potential indirect effect of climate change on public health is predicted to arise via climate-induced changes in aeroallergens, the primary risk factors for atopy and allergic asthma (D’Amato et al. 2010; Shea et al. 2008). The investigations in this dissertation relied on modeled NO2, which is a very good marker of within-city variability in TRAP. However, an important limitation arises from the fact that there were no personal 151  exposure measurements in this research. Relying on LUR models is likely to introduce error because these models explain between 76% of the variability in Vancouver to 81% in Winnipeg, leaving a non-negligible amount of unexplained variability and propagating error in the estimates subsequently used in further regressions for the epidemiological analyses.  Another limitation to note in this dissertation is the generally low level of air pollution. Throughout Canada, and certainly in the cities that were included in the present work, there has been a documented decline in NO2 measured at governmental air monitoring stations (Brook et al. 2014). Such low urban concentrations of NO2 are likely to hinder the ability to detect associations of modeled NO2 estimates with childhood asthma and allergies. However, studies of low-level exposures can be valuable in demonstrating that effects are harmful even at the low levels as shown in the present work, reinforcing the notion that “no safe concentration” has been identified.   6.1.3.2 Outcomes case definition Throughout the present dissertation, the ascertainment of both outcomes (childhood asthma and atopy) relied on an objective assessment and not on self-reported health status, which suffers from the threat of recall bias. However, there are different limitations related to each outcome that are best discussed separately, as follows. The assessment of atopy relied on an objective measure of sensitization by administering skin prick tests at each of the CHILD study center clinics. Skin allergy tests avoid the threat of recall bias and provide a reliable measure because the outcome is ascertained following a rigorous and consistent protocol across study sites. Furthermore, the study reported in this dissertation used a wide array of various allergens that capture the two main routes of exposure: ingestion and inhalation. However, criticism around this mode of case definition for atopy includes the lack of specificity when using a binary outcome based on an ad-152  hoc cut-off (Custovic et al. 2013). It has been shown that, when defined in this way, such positive allergy skin tests may indicate a positive allergic response without confirming clinical reactivity, where such defined atopic individuals have no evidence of asthma (Custovic et al. 1994). Instead, a continuous measure of wheal size would provide the opportunity to quantify the magnitude of the effect that air pollution has on allergic sensitization and to define “true” atopy in young children, which, when diagnosed among those with a history of wheezing, indicates a high risk of developing persistent asthma (Simpson et al. 2010; Sly et al. 2008). The definition of childhood asthma was based on administrative health records, which provides the advantage of avoiding recall bias. Asthma is a heterogeneous disease whose diagnosis is complex, even more so in the early years where individual wheezing episodes can be diagnosed wrongly as the development of asthma. Adding to this complexity is the fact that all asthma-related outcomes reported in this dissertation relied on physicians’ billing records for a visit related to asthma symptoms or diagnosis, rather than on a clinical record or an asthma-related medication prescription. Other issues include the validity of asthma case definitions using administrative data. A physician visit may be coded as asthma for suspected asthma in the absence of further testing for unusual symptoms that appear asthma-like, or in the absence of clear diagnostic criteria. However, these limitations are alleviated by the positive reports of multiple studies examining the ability of administrative data to identify asthma in Canada using visit or billing data collected as part of provincial universal healthcare system databases (Kozyrskyj et al. 2004; Prosser et al. 2008; To T et al. 2007). Furthermore, Chapter 5 confirms that linked administrative databases from mandatory healthcare insurance programs are suited for moving from incidence analyses to longer-term investigations of chronic episodic conditions. The limitation of defining asthma at one point in time, as performed in Chapter 4, is further confirmed by other Canadian studies of asthma recurrence in childhood. In Ontario, a similar administrative database was used to identify children who were diagnosed with asthma prior to the age of six and to track their asthma diagnoses until the age of 11. 153  In this study, asthma diagnosis was based on the same criteria as those used in the current dissertation. Children who continued to have asthma diagnoses past the age of six were considered to be persistent asthma cases. By age 12, half of the children were in remission. This finding highlights the advantage of using longitudinal and linked health databases to identify distinct long-term asthma groups for population-based studies.  6.2 Recommendations for future research  Given the asthma “epidemic” and the fact that the impact of exposure to many indoor and outdoor pollutants remains unresolved, the issue of an integrated assessment of environmental influences is of high importance because it may identify at-risk subpopulations (Gowers et al. 2012) as well as modifiable factors for which policy-makers can take appropriate measures.  The findings of this dissertation point to sources of exposure misclassification that may be present when relying on outdoor pollutant concentrations. These include infiltration and other exposures encountered in diverse microenvironments. Personal monitoring studies have shown relatively poor agreement with outdoor concentrations (Mölter et al. 2012a) and, more importantly, epidemiological studies using personal monitors have been unable to replicate associations (Mölter et al. 2013) found with outdoor pollutant estimates (Krämer et al. 2000; Pattenden et al. 2006). Could the associations between air pollution and pediatric respiratory outcomes found in previous studies based on outdoor concentrations be type I errors? Or were these outdoor concentrations a proxy for other environmental effects (e.g., moving from single to co-exposures related to built environment models)? Conflicting findings in past studies of air pollution epidemiology may have been due not only to exposure misclassification but also to poor outcome definitions and the lack of identification of susceptible subgroups. Were associations found in large population-based studies capturing a heterogeneous population encompassing several susceptible subgroups? A possible means of addressing this point in the 154  era of “big data” is by complementing incidence studies with the examination of the trajectory of this chronic heterogeneous outcome to encompass several endotypes of asthma. Two lines of research, detailed in the following sections, can be envisioned arising from these questions: 1) an exposure assessment research agenda and 2) cohort research with a focus on asthma endotype characterization. 6.2.1 Future directions for exposure assessment research Modeling indoor NO2 across micro-environments where a child typically spends known amounts of time has been shown to correlate well with personal measurements and to be implementable without prohibitive financial costs and overburdening participants in birth cohort studies (Mölter et al. 2012b). The next feasible step could be reconstructing a total exposure profile by time-averaging exposures in each micro-environment (Gulliver and Briggs 2005), especially because children do not display great variability in their time-activity patterns (Adgate et al. 2004). This approach holds promise to reduce exposure misclassification for cohort studies such as CHILD, especially in the presence of low levels of air pollution that require additional precision. With regard to exposure modeling, capturing infiltrated exposure can either be modeled based on particulate matter, or modeled using simplified chemical mass balance models, such as the INDAIR model used in the MAAS cohort study (Mölter et al. 2013). There are ongoing efforts to model infiltration efficiencies using easily obtainable predictors (Allen et al. 2012a) because infiltration may be an important source of exposure heterogeneity in epidemiologic studies of exposure to pollutants of outdoor origin. Similar models should be considered for use in Canadian settings, especially considering the efforts deployed in characterizing the physical environment in the CHILD study.  Hopanes can be found in different media, including house dust or window wipes and we explored their potential use in house dust. However, house dust as an exposure matrix has limitations related to the mode of accumulation. Several factors that may vary among study participants can affect the 155  concentrations of hopanes: cleaning practices and sampling surfaces (carpeted versus non-carpeted) play a significant role in the amount of chemicals that deposit inside the home. In addition, the metric of exposure for hopanes still lacks consensus, because hopanes can be measured in terms of the loading of chemicals (concentration normalized by surface area sampled) or expressed by the more traditional approach of normalized concentration to mass dust collected. The need to improve our understanding of the composition and deposition mechanisms in house dust is warranted. Work that was undertaken by Whitehead et al. with respect to PAHs in house dust should help guide future research (Whitehead et al. 2011, 2012, 2013). For example, it was shown that repeated measurements of house dust within study participants’ homes can improve the precision of PAH exposure estimates and limit attenuation bias. Investigation of hopanes using window wipes is another avenue for exploratory research, which may show less heterogeneity than house-settled dust.  LUR models provide high spatial resolution for exposure estimates of within-city variability, which has been shown to vary as much as that between cities. However, in light of urbanization trends and subsequent urban sprawl, methods are needed to expand the domain of LUR surfaces to include peri-urban/semi-rural areas. Hybrid models, which take advantage of the ability of dispersion modeling (Molter et al. 2010) or satellite data (Mao et al. 2012) to cover large geographical areas, represent a promising area of research.  In the case of population-based birth cohort studies such as BAQS, exposure assessment improvement can also be obtained through linkage with records from ministries of education. Cross-walk linkage between personal health identifiers and personal education numbers could therefore allow for the geocoding of schools and the estimation of exposure levels at these locations where children spend a significant amount of time. Time-weighted averages based on the home and school micro-environments would help shed more light on asthma analyses conducted on school-age participants. In addition, GIS tools can be used to approximate journey exposures, as shown by Gulliver and Briggs (Gulliver and 156  Briggs 2005). While this future research agenda will have limitations due to the coexistence of public and private school systems, it may shed more light on the disease etiology by reducing exposure misclassification for the school-age period.  6.2.2 Future directions for asthma and allergies research  As mentioned previously, prospective birth cohorts are the most powerful study design to assess the long-term effects of air pollution exposure on children’s respiratory health, taking into account that certain limitations as outlined above (see section 6.1.3) are addressed. In CHILD, the definition of childhood atopy should be defined as a continuous outcome to avoid a heuristic, albeit often-used, 2mm cut-off. The follow-up over time of allergy tests co-examined with other allergic phenotypes (e.g., eczema) can be used to predict different endotypes of asthma and to help understand the relationship between atopy and asthma. The lack of understanding of this relationship is related to the absence of a universally accepted definition of asthma. It is increasingly recognized that asthma actually represents several different diseases presenting with similar symptoms. Although sharing similar phenotypes, these distinct endotypes arise via unique and separate pathophysiological mechanisms that are likely to be associated with different environmental exposures and genetic markers (Lötvall et al. 2011). The most desirable aspect of endotyping asthma and atopy is that such endotypes would be expected to be more stable than their observable phenotypic features (e.g., wheezing for asthma or sensitization for atopy) (Anderson 2008; Custovic et al. 2013; Henderson 2014b). A more precise case definition that would result from this new paradigm of diagnosis would reduce the dilution of a true association with environmental exposures of interest. There has been a shift in the analysis of observed or collected symptom data in recent cohort research, mainly relying on new statistical approaches including latent class analysis (Henderson 2014b) with clusters replicated across different cohorts (Savenije et al. 2011), and on the unsupervised clustering 157  approach (PCA) to determine groups of phenotypes and define specific asthma endotypes in an unbiased way (Custovic et al. 2013; Lazic et al. 2013; Simpson et al. 2010). An improvement on the unsupervised approach to generate asthma classifications is adding the dimension of time to take into account longitudinal phenotype changes that may not be stable over time. We explored the use of Group Based Trajectory Modeling (GBTM) due to the fact that this tool offers promising research directions. For example, while our work examined time-invariant factors, GBTM can also incorporate time-varying covariates as well as multiple outcomes (Nagin and Odgers 2010), thus offering the possibility of examining co-morbidities (e.g., obesity) or behavioral outcomes (e.g., school absenteeism) that either overlap with or precede the development of asthma.  From a translational research perspective, such an approach can help inform the design of endotype-specific treatments. From a public health intervention standpoint, the ability to conduct informative population-based studies is dependent upon identifying different, longer-term asthma endotypes, such as persistent allergic asthma, that may be amenable to public health intervention via air pollution reduction measures or urban planning and design initiatives. Gilliland describes an additional facet to the exposure reduction paradigm by demonstrating that treatment of asthma phenotypes may be improved by the induction of enzymatic antioxidant defenses to pollutant insults to the immune system via dietary means, especially for individuals with high-risk genetic variants of key antioxidant enzymes (Gilliland 2009). The identification of a potentially protective role of greenness in this work is novel and important to note. Greenness may mitigate the effects of air pollution by improving air quality (Escobedo et al. 2011). Several other mechanisms by which the protective effect of green spaces may offset the risks of asthma and allergies, as outlined in the subsection 6.1.3.1, need to be explored further. Of particular interest is the finding by Mitchell and Popham (Mitchell and Popham 2008) that those exposed to the greenest environments also have the lowest levels of health inequality. Subsequent studies confirmed that the beneficial effects of green spaces on pregnancy outcomes seem to be more prominent among segments of 158  the population with low socio-economic status (Dadvand et al. 2012a). Because pregnancy outcomes have an impact on the newborn’s immune system, such studies need to be extended to pediatric outcomes. Physical environments that promote good health, such as green spaces, might be important to reduce socioeconomic health inequalities and therefore reduce the susceptibility of the lower-SES groups to the adverse effects of air pollution on respiratory health. Future studies of the effects of air pollution on pediatric chronic respiratory diseases would greatly benefit from the inclusion of other attributes of the built environment (e.g., green spaces, “blue” spaces, walkability) in order to inform policy-making decisions.  159  Bibliography Adgate JL, Church TR, Ryan AD, Ramachandran G, Fredrickson AL, Stock TH, et al. 2004. Outdoor, Indoor, and Personal Exposure to VOCs in Children. Environ. Health Perspect. 112: 1386–1392. Aguilera I, Pedersen M, Garcia-Esteban R, Ballester F, Basterrechea M, Esplugues A, et al. 2013. Early-life exposure to outdoor air pollution and respiratory health, ear infections, and eczema in infants from the INMA study. Environ. Health Perspect. 121:387–392; doi:10.1289/ehp.1205281. Allen RW, Adar SD, Avol E, Cohen M, Curl CL, Larson T, et al. 2012a. Modeling the Residential Infiltration of Outdoor PM2.5 in the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air). Environ. Health Perspect.; doi:10.1289/ehp.1104447. Allen RW, Adar SD, Avol E, Cohen M, Curl CL, Larson T, et al. 2012b. Modeling the Residential Infiltration of Outdoor PM2.5 in the Multi-Ethnic Study of Atherosclerosis and Air Pollution (MESA Air). Environ. Health Perspect. 120:824–830; doi:10.1289/ehp.1104447. Allen R, Wallace L, Larson T, Sheppard L, Liu L-JS. 2006. Evaluation of the recursive model approach for estimating particulate matter infiltration efficiencies using continuous light scattering data. J Expos Sci Env. Epidemiol 17: 468–477. Allen RW, Amram O, Wheeler AJ, Brauer M. 2011. The transferability of NO and NO2 land use regression models between cities and pollutants. Atmos. Environ. 45:369–378; doi:10.1016/j.atmosenv.2010.10.002. Anderson GP. 2008. Endotyping asthma: new insights into key pathogenic mechanisms in a complex, heterogeneous disease. The Lancet 372:1107–1119; doi:10.1016/S0140-6736(08)61452-X. Anderson Hr, Favarato G, Atkinson RW. 2013a. Long-term exposure to air pollution and the incidence of asthma: meta-analysis of cohort studies. Air Qual. Atmosphere Health 6:47–56; doi:http://dx.doi.org/10.1605/01.301-0022014439.2013. Anderson Hr, Favarato G, Atkinson RW. 2013b. Long-term exposure to outdoor air pollution and the prevalence of asthma: meta-analysis of multi-community prevalence studies. Air Qual. Atmosphere Health 6:57–68; doi:http://dx.doi.org/10.1007/s11869-011-0145-4. Anderson HR, Ruggles R, Pandey KD, Kapetanakis V, Brunekreef B, Lai CKW, et al. 2010. Ambient particulate pollution and the world-wide prevalence of asthma, rhinoconjunctivitis and eczema in children: Phase One of the International Study of Asthma and Allergies in Childhood (ISAAC). Occup. Environ. Med. 67:293–300; doi:10.1136/oem.2009.048785. Arbes SJ, Gergen PJ, Vaughn B, Zeldin DC. 2007. Asthma cases attributable to atopy: results from the Third National Health and Nutrition Examination Survey. J. Allergy Clin. Immunol. 120:1139–1145; doi:10.1016/j.jaci.2007.07.056. 160  Armstrong BG. 1998. Effect of measurement error on epidemiological studies of environmental and occupational exposures. Occup. Environ. Med. 55: 651–656. Arrandale VH, Brauer M, Brook JR, Brunekreef B, Gold DR, London SJ, et al. 2010. Exposure Assessment in Cohort Studies of Childhood Asthma. Environ. Health Perspect.; doi:10.1289/ehp.1002267. Arshad SH. 2010. Does exposure to indoor allergens contribute to the development of asthma and allergy? Curr. Allergy Asthma Rep. 10: 49–55. Asher MI, Montefort S, Björkstén B, Lai CKW, Strachan DP, Weiland SK, et al. 2006. Worldwide time trends in the prevalence of symptoms of asthma, allergic rhinoconjunctivitis, and eczema in childhood: ISAAC Phases One and Three repeat multicountry cross-sectional surveys. Lancet 368: 733–743. Beckerman B, Jerrett M, Brook JR, Verma DK, Arain MA, Finkelstein MM. 2008. Correlation of nitrogen dioxide with other traffic pollutants near a major expressway. Atmos. Environ. 42:275–290; doi:16/j.atmosenv.2007.09.042. Beelen R, Hoek G, van den Brandt PA, Goldbohm RA, Fischer P, Schouten LJ, et al. 2008. Long-Term Exposure to Traffic-Related Air Pollution and Lung Cancer Risk. Epidemiology 19:702–710; doi:10.1097/EDE.0b013e318181b3ca. Beggs PJ. 2004. Impacts of climate change on aeroallergens: past and future. Clin. Exp. Allergy 34:1507–1513; doi:10.1111/j.1365-2222.2004.02061.x. Belgrave DCM, Custovic A, Simpson A. 2013. Characterizing wheeze phenotypes to identify endotypes of childhood asthma, and the implications for future management. Expert Rev. Clin. Immunol. 9:921–936; doi:10.1586/1744666X.2013.836450. Bisgaard H, Bønnelykke K. 2010. Long-term studies of the natural history of asthma in childhood. J. Allergy Clin. Immunol. 126:187–197; quiz 198–199; doi:10.1016/j.jaci.2010.07.011. Bisgaard H, Simpson A, Palmer CN., Bønnelykke K, Mclean I, Mukhopadhyay S, et al. 2008. Gene-Environment Interaction in the Onset of Eczema in Infancy: Filaggrin Loss-of-Function Mutations Enhanced by Neonatal Cat Exposure. PLoS Med 5:e131; doi:10.1371/journal.pmed.0050131. Blais L, Lemière C, Menzies D, Berbiche D. 2006. Validity of asthma diagnoses recorded in the Medical Services database of Quebec. Pharmacoepidemiol. Drug Saf. 15:245–252; doi:10.1002/pds.1202. Bousquet J, Bousquet PJ, Godard P, Daures J-P. 2005. The public health implications of asthma. Bull. World Health Organ. 83: 548–554. Bråbäck L, Forsberg B. 2009. Does traffic exhaust contribute to the development of asthma and allergic sensitization in children: findings from recent cohort studies. Environ. Health Glob. Access Sci. Source 8:17; doi:10.1186/1476-069X-8-17. 161  Brauer M, Hoek G, Smit HA, de Jongste JC, Gerritsen J, Postma DS, et al. 2007a. Air pollution and development of asthma, allergy and infections in a birth cohort. Eur. Respir. J. 29:879–888; doi:10.1183/09031936.00083406. Brauer M, Hoek G, Smit HA, Jongste JC de, Gerritsen J, Postma DS, et al. 2007b. Air pollution and development of asthma, allergy and infections in a birth cohort. Eur. Respir. J. 29:879–888; doi:10.1183/09031936.00083406. Brauer M, Hoek G, Van Vliet P, Meliefste K, Fischer PH, Wijga A, et al. 2002. Air pollution from traffic and the development of respiratory infections and asthmatic and allergic symptoms in children. Am. J. Respir. Crit. Care Med. 166: 1092–1098. Brauer M, Lencar C, Tamburic L, Koehoorn M, Demers P, Karr C. 2008. A Cohort Study of Traffic-Related Air Pollution Impacts on Birth Outcomes. Environ. Health Perspect. 116:680–686; doi:10.1289/ehp.10952. Brauer M, Reynolds C, Hystad P. 2012. Traffic-related air pollution and health : a Canadian perspective on scientific evidence and potential exposure-mitigation strategies. Briggs D. 2005. The Role of Gis: Coping With Space (And Time) in Air Pollution Exposure Assessment. J. Toxicol. Environ. Health Part Curr. Issues 68:1243; doi:10.1080/15287390590936094. Briggs DJ, Hoogh C de, Gulliver J, Wills J, Elliott P, Kingham S, et al. 2000. A regression-based method for mapping traffic-related air pollution: application and testing in four contrasting urban environments. Sci. Total Environ. 253:151 – 167; doi:DOI: 10.1016/S0048-9697(00)00429-0. British Columbia (BC) Vital Statistics Agency [creator]. 2009a. Vital Statistics Births. V2. Population Data BC [publisher]. Data Extract. BC Vital Statistics Agency (2009). http://www.popdata.bc.ca/data British Columbia (BC) Vital Statistics Agency. 2009b. Vital Statistics Deaths. V2. Population Data BC [publisher]. Data Extract. BC Vital Statistics Agency (2009). http://www.popdata.bc.ca/data British Columbia Ministry of Health. 2009a. Discharge Abstract Database (Hospital Separations) V2. Population Data BC [publisher]. Data Extract. MOH (2009). http://www.popdata.bc.ca/data British Columbia Ministry of Health. 2009b. Medical Services Plan (MSP) Payment Information File. V2. Population Data BC [publisher]. Data Extract. MOH (2009). http://www.popdata.bc.ca/data Brook JR, Dann TF, Galarneau E, Herod D, Charland JP. 2014. The State of Air Quality in Canada: National Patterns. In Air Quality Management (E. Taylor and A. McMillaneds. ), pp. 43–67, Springer Netherlands. Brunekreef B, Janssen NA, de Hartog J, Harssema H, Knape M, van Vliet P. 1997. Air pollution from truck traffic and lung function in children living near motorways. Epidemiol. Camb. Mass 8: 298–303. 162  Brunekreef B, Stewart AW, Anderson HR, Lai CKW, Strachan DP, Pearce N. 2009. Self-reported truck traffic on the street of residence and symptoms of asthma and allergic disease: a global relationship in ISAAC phase 3. Environ. Health Perspect. 117: 1791–1798. California, EPA. 1997. Health Effects of Exposure to Environmental Tobacco Smoke. Canadian Institute for Health Information (CIHI). 2011. In Due Time: Why Maternal Age Matters. Available at: https://secure.cihi.ca/free_products/AIB_InDueTime_WhyMaternalAgeMatters_E.pdf Cariñanos P, Casares-Porcel M. 2011. Urban green zones and related pollen allergy: A review. Some guidelines for designing spaces with low allergy impact. Landsc. Urban Plan. 101:205–214; doi:10.1016/j.landurbplan.2011.03.006. Carlsen K-H, Carlsen KCL. 2008. Respiratory effects of tobacco smoking on infants and young children. Paediatr. Respir. Rev. 9:11–19; quiz 19–20; doi:10.1016/j.prrv.2007.11.007. Celedon JC, Milton DK, Ramsey CD, Litonjua AA, Ryan L, Platts-Mills TA, et al. 2007. Exposure to dust mite allergen and endotoxin in early life and asthma and atopy in childhood. J. Allergy Clin. Immunol. 120: 144–149. Chen Q, Just AC, Miller RL, Perzanowski MS, Goldstein IF, Perera FP, et al. 2012. Using latent class growth analysis to identify childhood wheeze phenotypes in an urban birth cohort. Ann. Allergy. Asthma. Immunol. 108:311–315.e1; doi:10.1016/j.anai.2012.02.016. Clark NA, Allen RW, Hystad P, Wallace L, Dell SD, Foty R, et al. 2010a. Exploring Variation and Predictors of Residential Fine Particulate Matter Infiltration. Int. J. Environ. Res. Public. Health 7:3211–3224; doi:10.3390/ijerph7083211. Clark NA, Demers PA, Karr CJ, Koehoorn M, Lencar C, Tamburic L, et al. 2010b. Effect of early life exposure to air pollution on development of childhood asthma. Environ. Health Perspect. 118:284–290; doi:10.1289/ehp.0900916. Corren J. 2013. Asthma Phenotypes and Endotypes: An Evolving Paradigm for Classification. Discov. Med. 15: 243–249. Custovic A, Arifhodzic N, Robinson A, Woodcock A. 1994. EXercise testing revisited. the response to exercise in normal and atopic children. Chest 105:1127–1132; doi:10.1378/chest.105.4.1127. Custovic A, Lazic N, Simpson A. 2013. Pediatric asthma and development of atopy: Curr. Opin. Allergy Clin. Immunol. 13:173–180; doi:10.1097/ACI.0b013e32835e82b6. Dabek-Zlotorzynska E, Dann TF, Kalyani Martinelango P, Celo V, Brook JR, Mathieu D, et al. 2011. Canadian National Air Pollution Surveillance (NAPS) PM2.5 speciation program: Methodology and PM2.5 chemical composition for the years 2003–2008. Atmos. Environ. 45:673–686; doi:10.1016/j.atmosenv.2010.10.024. 163  Dadvand P, de Nazelle A, Figueras F, Basagaña X, Su J, Amoly E, et al. 2012a. Green space, health inequality and pregnancy. Environ. Int. 40:110–115; doi:10.1016/j.envint.2011.07.004. Dadvand P, Sunyer J, Basagana X, Ballester F, Lertxundi A, Fernandez-Somoano A, et al. 2012b. Surrounding Greenness and Pregnancy Outcomes in Four Spanish Birth Cohorts. Environ. Health Perspect. 120:1481–1487; doi:10.1289/ehp.1205244. Dadvand P, Villanueva CM, Font-Ribera L, Martinez D, Basagaña X, Belmonte J, et al. 2014. Risks and Benefits of Green Spaces for Children: A Cross-Sectional Study of Associations with Sedentary Behavior, Obesity, Asthma, and Allergy. Environ. Health Perspect.; doi:10.1289/ehp.1308038. Dales RE, Cakmak S, Judek S, Dann T, Coates F, Brook JR, et al. 2004. Influence of outdoor aeroallergens on hospitalization for asthma in Canada. J. Allergy Clin. Immunol. 113:303–306; doi:10.1016/j.jaci.2003.11.016. Dales R, Miller D, Ruest K, Guay M, Judek S. 2006. Airborne endotoxin is associated with respiratory illness in the first 2 years of life. Environ. Health Perspect. 114: 610–614. Dales R, Wheeler AJ, Mahmud M, Frescura A-M, Liu L. 2009. The Influence of Neighborhood Roadways on Respiratory Symptoms Among Elementary Schoolchildren. J. Occup. Environ. Med. 51:654–660; doi:10.1097/JOM.0b013e3181a0363c. D’Amato G, Cecchi L. 2008. Effects of climate change on environmental factors in respiratory allergic diseases. Clin. Exp. Allergy J. Br. Soc. Allergy Clin. Immunol. 38:1264–1274; doi:10.1111/j.1365-2222.2008.03033.x. D’Amato G, Cecchi L, Bonini S, Nunes C, Annesi-Maesano I, Behrendt H, et al. 2007. Allergenic pollen and pollen allergy in Europe. Allergy 62:976–990; doi:10.1111/j.1398-9995.2007.01393.x. D’Amato G, Cecchi L, D’Amato M, Liccardi G. 2010. Urban air pollution and climate change as environmental risk factors of respiratory allergy: an update. J. Investig. Allergol. Clin. Immunol. 20: 95. D’Amato G, Liccardi G, D’Amato M, Cazzola M. 2002. Outdoor air pollution, climatic changes and allergic bronchial asthma. Eur. Respir. J. Off. J. Eur. Soc. Clin. Respir. Physiol. 20: 763–776. Delfino RJ, Staimer N, Tjoa T, Arhami M, Polidori A, Gillen DL, et al. 2010. Association of Biomarkers of Systemic Inflammation with Organic Components and Source Tracers in Quasi-Ultrafine Particles. Environ Health Perspect 118:756–762; doi:10.1289/ehp.0901407. Dell SD, Foty RG, Gilbert NL, Jerrett M, To T, Walter SD, et al. 2010. Asthma and allergic disease prevalence in a diverse sample of Toronto school children: Results from the Toronto Child Health Evaluation Questionnaire (T-CHEQ) Study. Can. Respir. J. J. Can. Thorac. Soc. 17: e1–e6. Devereux G, Barker RN, Seaton A. 2002. Antenatal determinants of neonatal immune responses to allergens. Clin. Exp. Allergy J. Br. Soc. Allergy Clin. Immunol. 32: 43–50. 164  Dietert RR, Etzel RA, Chen D, Halonen M, Holladay SD, Jarabek AM, et al. 2000. Workshop to identify critical windows of exposure for children’s health: immune and respiratory systems work group summary. Environ. Health Perspect. 108: 483–490. Dietert RR, Zelikoff JT. 2008. Early-life environment, developmental immunotoxicology, and the risk of pediatric allergic disease including asthma. Birth Defects Res. B Dev. Reprod. Toxicol. 83: 547–560. Dijkema MB, Gehring U, van Strien RT, van der Zee SC, Fischer P, Hoek G, et al. 2011. A Comparison of Different Approaches to Estimate Small-Scale Spatial Variation in Outdoor NO2 Concentrations. Environ. Health Perspect. 119:670–675; doi:10.1289/ehp.0901818. Dizier MH, Bouzigon E, Guilloud-Bataille M, Siroux V, Lemainque A, Boland A, et al. 2007. Evidence for gene x smoking exposure interactions in a genome-wide linkage screen of asthma and bronchial hyper-responsiveness in EGEA families. Eur. J. Hum. Genet. EJHG 15: 810–815. Dominici F, Cefalu M. 2015. Confounding Adjustment and Exposure Prediction in Environmental Epidemiology: Additional Insights. Epidemiology 26:e28; doi:10.1097/EDE.0000000000000231. Douwes J, van Strien R, Doekes G, Smit J, Kerkhof M, Gerritsen J, et al. 2006. Does early indoor microbial exposure reduce the risk of asthma? The Prevention and Incidence of Asthma and Mite Allergy birth cohort study. J. Allergy Clin. Immunol. 117:1067–1073; doi:10.1016/j.jaci.2006.02.002. Dzhambov AM, Dimitrova DD, Dimitrakova ED. 2014. Association between residential greenness and birth weight: Systematic review and meta-analysis. Urban Forestry & Urban Greening 13:621–629; doi:10.1016/j.ufug.2014.09.004. Eder W, Ege MJ, von Mutius E. 2006. The asthma epidemic. N. Engl. J. Med. 355:2226–2235; doi:10.1056/NEJMra054308. Egeghy PP, Quackenboss JJ, Catlin S, Ryan PB. 2004. Determinants of temporal variability in NHEXAS-Maryland environmental concentrations, exposures, and biomarkers. J Expo Anal Env. Epidemiol 15: 388–397. Ege MJ, Mayer M, Normand A-C, Genuneit J, Cookson WOCM, Braun-Fahrländer C, et al. 2011. Exposure to environmental microorganisms and childhood asthma. N. Engl. J. Med. 364:701–709; doi:10.1056/NEJMoa1007302. Escobedo FJ, Kroeger T, Wagner JE. 2011. Urban forests and pollution mitigation: Analyzing ecosystem services and disservices. Environ. Pollut. 159:2078–2087; doi:10.1016/j.envpol.2011.01.010. Esplugues A, Ballester F, Estarlich M, Llop S, Fuentes-Leonarte V, Mantilla E, et al. 2011. Outdoor, but not indoor, nitrogen dioxide exposure is associated with persistent cough during the first year of life. Sci. Total Environ. 409:4667–4673; doi:10.1016/j.scitotenv.2011.08.007. 165  Favarato G, Anderson HR, Atkinson R, Fuller G, Mills I, Walton H. 2014. Traffic-related pollution and asthma prevalence in children. Quantification of associations with nitrogen dioxide. Air Qual. Atmosphere Health 7:459–466; doi:10.1007/s11869-014-0265-8. Fisk WJ, Lei‐Gomez Q, Mendell MJ. 2007. Meta‐analyses of the associations of respiratory health effects with dampness and mold in homes. Indoor Air 17:284–296; doi:10.1111/j.1600-0668.2007.00475.x. Gan WQ, Tamburic L, Davies HW, Demers PA, Koehoorn M, Brauer M. 2010. Changes in residential proximity to road traffic and the risk of death from coronary heart disease. Epidemiol. Camb. Mass 21:642–649; doi:10.1097/EDE.0b013e3181e89f19. Garner R, Kohen D. 2008. Changes in the prevalence of asthma among Canadian children. Health Rep. 19: 45–50. Gasana J, Dillikar D, Mendy A, Forno E, Ramos Vieira E. 2012. Motor vehicle air pollution and asthma in children: a meta-analysis. Environ. Res. 117:36–45; doi:10.1016/j.envres.2012.05.001. Gauderman WJ, Avol E, Lurmann F, Kuenzli N, Gilliland F, Peters J, et al. 2005. Childhood asthma and exposure to traffic and nitrogen dioxide. Epidemiol. Camb. Mass 16: 737–743. Gehring U, Cyrys J, Sedlmeir G, Brunekreef B, Bellander T, Fischer P, et al. 2002. Traffic-related air pollution and respiratory health during the first 2 yrs of life. Eur. Respir. J. 19:690 –698; doi:10.1183/09031936.02.01182001. Gehring U, Pattenden S, Slachtova H, Antova T, Braun-Fahrländer C, Fabianova E, et al. 2006. Parental education and children’s respiratory and allergic symptoms in the Pollution and the Young (PATY) study. Eur. Respir. J. 27:95–107; doi:10.1183/09031936.06.00017205. Gehring U, Wijga AH, Brauer M, Fischer P, de Jongste JC, Kerkhof M, et al. 2010a. Traffic-related air pollution and the development of asthma and allergies during the first 8 years of life. Am. J. Respir. Crit. Care Med. 181:596–603; doi:10.1164/rccm.200906-0858OC. Gehring U, Wijga AH, Brauer M, Fischer P, de Jongste JC, Kerkhof M, et al. 2010b. Traffic-related Air Pollution and the Development of Asthma and Allergies during the First 8 Years of Life. Am. J. Respir. Crit. Care Med. 181:596–603; doi:10.1164/rccm.200906-0858OC. Gilbert NL, Goldberg MS, Beckerman B, Brook JR, Jerrett M. 2005. Assessing spatial variability of ambient nitrogen dioxide in Montreal, Canada, with a land-use regression model. J. Air Waste Manag. Assoc. 55: 1059–1063. Gilliland FD. 2009. Outdoor Air Pollution, Genetic Susceptibility, and Asthma Management: Opportunities for Intervention to Reduce the Burden of Asthma. PEDIATRICS 123:S168–S173; doi:10.1542/peds.2008-2233G. 166  Gilliland FD, Li YF, Peters JM. 2001. Effects of maternal smoking during pregnancy and environmental tobacco smoke on asthma and wheezing in children. Am. J. Respir. Crit. Care Med. 163:429–436; doi:10.1164/ajrccm.163.2.2006009. Government of Canada EC. 2007. CESI Home Page - Environmental Indicators - Environment Canada. Available: http://ec.gc.ca/indicateurs-indicators/default.asp?lang=En [accessed 7 July 2015].  Gowers AM, Cullinan P, Ayres JG, Anderson HR, Strachan DP, Holgate ST, et al. 2012. Does outdoor air pollution induce new cases of asthma? Biological plausibility and evidence; a review. Respirol. Carlton Vic 17:887–898; doi:10.1111/j.1440-1843.2012.02195.x. Graham LA, Tong A, Poole G, Ding L, Ke F, Wang D, et al. 2010. A comparison of direct thermal desorption with solvent extraction for gas chromatography-mass spectrometry analysis of semivolatile organic compounds in diesel particulate matter. Int. J. Environ. Anal. Chem. 90:511–534; doi:10.1080/03067310903108352. Granell R, Sterne JAC, Henderson J. 2012. Associations of different phenotypes of wheezing illness in early childhood with environmental variables implicated in the aetiology of asthma. PloS One 7:e48359; doi:10.1371/journal.pone.0048359. Greater Vancouver Regional District. Resources Metro Vancouver 2040: Shaping our Future. Available: http://www.metrovancouver.org/services/regional-planning/metro-vancouver-2040/resources/Pages/default.aspx [accessed 3 April 2015]. Gruzieva O, Bellander T, Eneroth K, Kull I, Melén E, Nordling E, et al. 2012. Traffic-related air pollution and development of allergic sensitization in children during the first 8 years of life. J. Allergy Clin. Immunol. 129:240–246; doi:10.1016/j.jaci.2011.11.001. Gruzieva O, Bergström A, Hulchiy O, Kull I, Lind T, Melén E, et al. 2013. Exposure to air pollution from traffic and childhood asthma until 12 years of age. Epidemiol. Camb. Mass 24:54–61; doi:10.1097/EDE.0b013e318276c1ea. Gruzieva O, Gehring U, Aalberse R, Agius R, Beelen R, Behrendt H, et al. 2014. Meta-analysis of air pollution exposure association with allergic sensitization in European birth cohorts. J. Allergy Clin. Immunol. 133:767–776.e7; doi:10.1016/j.jaci.2013.07.048. Guarnieri M, Balmes JR. 2014. Outdoor air pollution and asthma. The Lancet 383:1581–1592; doi:10.1016/S0140-6736(14)60617-6. Gulliver J, Briggs DJ. 2005. Time–space modeling of journey-time exposure to traffic-related air pollution using GIS. Environ. Res. 97:10–25; doi:10.1016/j.envres.2004.05.002. Gulliver J, de Hoogh K, Fecht D, Vienneau D, Briggs D. 2011. Comparative assessment of GIS-based methods and metrics for estimating long-term exposures to air pollution. Atmos. Environ. 45:7072–7080; doi:10.1016/j.atmosenv.2011.09.042. 167  Haahtela T, Holgate S, Pawankar R, Akdis CA, Benjaponpitak S, Caraballo L, et al. 2013. The biodiversity hypothesis and allergic disease: world allergy organization position statement. World Allergy Organ. J. 6:3; doi:10.1186/1939-4551-6-3. Hartig T, Mitchell R, de Vries S, Frumkin H. 2014. Nature and Health. Annu. Rev. Public Health 35:207–228; doi:10.1146/annurev-publhealth-032013-182443. Health Effects Institute. 2010. Traffic Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Heinrich J. 2011. Influence of indoor factors in dwellings on the development of childhood asthma. International Journal of Hygiene and Environmental Health 214:1–25; doi:10.1016/j.ijheh.2010.08.009. Heinrich J, Wichmann H-E. 2004. Traffic related pollutants in Europe and their effect on allergic disease. Curr. Opin. Allergy Clin. Immunol. Oct. 2004 4: 341–348. Henderson AJ. 2014a. Childhood asthma phenotypes in the twenty-first century. Breathe 10:100–108; doi:10.1183/20734735.014613. Henderson AJ. 2014b. Childhood asthma phenotypes in the twenty-first century. Breathe 10:100–108; doi:10.1183/20734735.014613. Henderson J, Granell R, Heron J, Sherriff A, Simpson A, Woodcock A, et al. 2008. Associations of wheezing phenotypes in the first 6 years of life with atopy, lung function and airway responsiveness in mid-childhood. Thorax 63:974–980; doi:10.1136/thx.2007.093187. Henderson SB, Beckerman B, Jerrett M, Brauer M. 2007. Application of land use regression to estimate long-term concentrations of traffic-related nitrogen oxides and fine particulate matter. Environ. Sci. Technol. 41: 2422–2428. Hoek G, Beelen R, de Hoogh K, Vienneau D, Gulliver J, Fischer P, et al. 2008. A review of land-use regression models to assess spatial variation of outdoor air pollution. Atmos. Environ. 42:7561–7578; doi:10.1016/j.atmosenv.2008.05.057. Huang X-F, Chen D-L, Lan Z-J, Feng N, He L-Y, Yu G-H, et al. 2012. Characterization of organic aerosol in fine particles in a mega-city of South China: Molecular composition, seasonal variation, and size distribution. Atmospheric Res. 114–115:28–37; doi:10.1016/j.atmosres.2012.05.019. Hu PS. 2005. Summary of Travel Trends 2001 National Household Travel Survey. Huss K, Jr NFA, Eggleston PA, Dawson C, Natta MLV, Hamilton RG. 2001. House dust mite and cockroach exposure are strong risk factors for positive allergy skin test responses in the Childhood Asthma Management Program. J. Allergy Clin. Immunol. 107: 48–54. 168  Hystad P, Davies HW, Frank L, Van Loon J, Gehring U, Tamburic L, et al. 2014. Residential greenness and birth outcomes: evaluating the influence of spatially correlated built-environment factors. Environ. Health Perspect. 122:1095–1102; doi:10.1289/ehp.1308049. Hystad PU, Setton EM, Allen RW, Keller PC, Brauer M. 2009. Modeling residential fine particulate matter infiltration for exposure assessment. J. Expo. Sci. Environ. Epidemiol. 19: 570–579. Iossifova YY, Reponen T, Bernstein DI, Levin L, Kalra H, Campo P, et al. 2007. House dust (1–3)‐β‐d‐glucan and wheezing in infants. Allergy 62:504–513; doi:10.1111/j.1398-9995.2007.01340.x. Ismaila AS, Sayani AP, Marin M, Su Z. 2013. Clinical, economic, and humanistic burden of asthma in Canada: a systematic review. BMC Pulm. Med. 13:70; doi:10.1186/1471-2466-13-70. Janssen NAH, Brunekreef B, van Vliet P, Aarts F, Meliefste K, Harssema H, et al. 2003. The relationship between air pollution from heavy traffic and allergic sensitization, bronchial hyperresponsiveness, and respiratory symptoms in Dutch schoolchildren. Environ. Health Perspect. 111: 1512–1518. Jedrychowski W, Galas A, Pac A, Flak E, Camman D, Rauh V, et al. 2005. Prenatal ambient air exposure to polycyclic aromatic hydrocarbons and the occurrence of respiratory symptoms over the first year of life. Eur. J. Epidemiol. 20:775–782; doi:10.1007/s10654-005-1048-1. Jerrett M, Arain MA, Kanaroglou P, Beckerman B, Crouse D, Gilbert NL, et al. 2007. Modeling the intraurban variability of ambient traffic pollution in Toronto, Canada. J. Toxicol. Environ. Health A 70: 200–212. Jerrett M, Shankardass K, Berhane K, Gauderman WJ, Künzli N, Avol E, et al. 2008. Traffic-Related Air Pollution and Asthma Onset in Children: A Prospective Cohort Study with Individual Exposure Measurement. Environ. Health Perspect. 116:1433–1438; doi:10.1289/ehp.10968. Jiang R, Jones MJ, Sava F, Kobor MS, Carlsten C. 2014. Short-term diesel exhaust inhalation in a controlled human crossover study is associated with changes in DNA methylation of circulating mononuclear cells in asthmatics. Part. Fibre Toxicol. 11; doi:10.1186/s12989-014-0071-3. Jobe AH, Tibboel D. 2014. Update in Pediatric Lung Disease 2013. Am. J. Respir. Crit. Care Med. 189:1031–1036; doi:10.1164/rccm.201402-0230UP. Karner AA, Eisinger DS, Niemeier DA. 2010. Near-Roadway Air Quality: Synthesizing the Findings from Real-World Data. Environ. Sci. Technol. 44:5334–5344; doi:10.1021/es100008x. Karr CJ, Demers PA, Koehoorn MW, Lencar CC, Tamburic L, Brauer M. 2009. Influence of ambient air pollutant sources on clinical encounters for infant bronchiolitis. Am. J. Respir. Crit. Care Med. 180:995–1001; doi:10.1164/rccm.200901-0117OC. Kaur S, Nieuwenhuijsen MJ. 2009. Determinants of Personal Exposure to PM2.5, Ultrafine Particle Counts, and CO in a Transport Microenvironment. Environ. Sci. Technol. 43:4737–4743; doi:10.1021/es803199z. 169  Kerkhof M, Postma DS, Brunekreef B, Reijmerink NE, Wijga AH, de Jongste JC, et al. 2010. Toll-like receptor 2 and 4 genes influence susceptibility to adverse effects of traffic-related air pollution on childhood asthma. Thorax 65:690–697; doi:10.1136/thx.2009.119636. Klepeis NE, Nelson WC, Ott WR, Robinson JP, Tsang AM, Switzer P, et al. 2001. The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Anal. Environ. Epidemiol. 11: 231–252. Kozyrskyj AL, Mustard CA, Becker AB. 2004. Identifying children with persistent asthma from health care administrative records. Can. Respir. J. J. Can. Thorac. Soc. 11: 141–145. Krämer U, Koch T, Ranft U, Ring J, Behrendt H. 2000. Traffic-related air pollution is associated with atopy in children living in urban areas. Epidemiology 11: 64–70. Krämer U, Sugiri D, Ranft U, Krutmann J, von Berg A, Berdel D, et al. 2009. Eczema, respiratory allergies, and traffic-related air pollution in birth cohorts from small-town areas. J. Dermatol. Sci. 56:99–105; doi:10.1016/j.jdermsci.2009.07.014. Labrèche F, Kosatsky T, Przybysz R. 2008. Childhood asthma surveillance using administrative data: consistency between medical billing and hospital discharge diagnoses. Can. Respir. J. J. Can. Thorac. Soc. 15: 188–192. Larson T, Henderson SB, Brauer M. 2009. Mobile monitoring of particle light absorption coefficient in an urban area as a basis for land use regression. Environ. Sci. Technol. 43: 4672–4678. Lau S, Illi S, Platts-Mills TA, Riposo D, Nickel R, Gruber C, et al. 2005. Longitudinal study on the relationship between cat allergen and endotoxin exposure, sensitization, cat-specific IgG and development of asthma in childhood–report of the German Multicentre Allergy Study (MAS 90). Allergy 60: 766–773. Lazic N, Roberts G, Custovic A, Belgrave D, Bishop CM, Winn J, et al. 2013. Multiple atopy phenotypes and their associations with asthma: similar findings from two birth cohorts. Allergy 68:764–770; doi:10.1111/all.12134. Lee ACK, Maheswaran R. 2011. The health benefits of urban green spaces: a review of the evidence. J. Public Health 33:212–222; doi:10.1093/pubmed/fdq068. Leech JA, Nelson WC, Burnett RT, Aaron S, Raizenne ME. 2002. It’s about time: A comparison of Canadian and American time-activity patterns. J Expo Anal Env. Epidemiol 12: 427–432. Lemanske Jr. RF, Busse WW. 2010. Asthma: Clinical expression and molecular mechanisms. J. Allergy Clin. Immunol. 125:S95–S102; doi:10.1016/j.jaci.2009.10.047. Lin L, Lee ML, Eatough DJ. 2010. Review of Recent Advances in Detection of Organic Markers in Fine Particulate Matter and Their Use for Source Apportionment. J. Air Waste Manag. Assoc. 60:3–25; doi:10.3155/1047-3289.60.1.3. 170  Lioy PJ, Freeman NCG, Millette JR. 2002. Dust: a metric for use in residential and building exposure assessment and source characterization. Environ. Health Perspect. 110: 969–983. Lix LM, Yogendran MS, Shaw SY, Burchill C, Metge C, Bond R. 2008. Population-based data sources for chronic disease surveillance. Chronic Dis. Can. 29: 31–38. Loan R, Siebers R, Fitzharris P, Crane J. 2003. House dust-mite allergen and cat allergen variability within carpeted living room floors in domestic dwellings. Indoor Air 13: 232–236. London SJ. 2007. Gene-Air Pollution Interactions in Asthma. Proc Am Thorac Soc 4:217–220; doi:10.1513/pats.200701-031AW. Loo CKJ, Foty RG, Wheeler AJ, Miller JD, Evans G, Stieb DM, et al. 2010. Do questions reflecting indoor air pollutant exposure from a questionnaire predict direct measure of exposure in owner-occupied houses? Int. J. Environ. Res. Public. Health 7:3270–3297; doi:10.3390/ijerph7083270. Los H, Postmus PE, Boomsma DI. 2001. Asthma Genetics and Intermediate Phenotypes: A Review From Twin Studies. Twin Research and Human Genetics 4:81–93; doi:10.1375/twin.4.2.81. Lötvall J, Akdis CA, Bacharier LB, Bjermer L, Casale TB, Custovic A, et al. 2011. Asthma endotypes: A new approach to classification of disease entities within the asthma syndrome. J. Allergy Clin. Immunol. 127:355–360; doi:10.1016/j.jaci.2010.11.037. Lovasi GS, O’Neil-Dunne JPM, Lu JWT, Sheehan D, Perzanowski MS, MacFaden S, et al. 2013. Urban Tree Canopy and Asthma, Wheeze, Rhinitis, and Allergic Sensitization to Tree Pollen in a New York City Birth Cohort. Environ. Health Perspect.; doi:10.1289/ehp.1205513. Lovasi GS, Quinn JW, Neckerman KM, Perzanowski MS, Rundle A. 2008. Children living in areas with more street trees have lower prevalence of asthma. J. Epidemiol. Community Health 62:647–649; doi:10.1136/jech.2007.071894. MacIntyre EA, Brauer M, Melén E, Bauer CP, Bauer M, Berdel D, et al. 2014. GSTP1 and TNF Gene Variants and Associations between Air Pollution and Incident Childhood Asthma: The Traffic, Asthma and Genetics (TAG) Study. Environ. Health Perspect. 122:418–424; doi:10.1289/ehp.1307459. MacIntyre EA, Karr CJ, Koehoorn M, Demers PA, Tamburic L, Lencar C, et al. 2011. Residential air pollution and otitis media during the first two years of life. Epidemiol. Camb. Mass 22:81–89; doi:10.1097/EDE.0b013e3181fdb60f. Maddox L, Schwartz DA. 2002. The pathophysiology of asthma. Annu. Rev. Med. 53:477–498; doi:10.1146/annurev.med.53.082901.103921. Mahler BJ, Metre PCV, Wilson JT, Musgrove M, Burbank TL, Ennis TE, et al. 2010. Coal-Tar-Based Parking Lot Sealcoat: An Unrecognized Source of PAH to Settled House Dust. Env. Sci Technol 44:894–900; doi:10.1021/es902533r. 171  Mao L, Qiu Y, Kusano C, Xu X. 2012. Predicting regional space-time variation of PM2.5 with land-use regression model and MODIS data. Environ. Sci. Pollut. Res. Int. 19:128–138; doi:10.1007/s11356-011-0546-9. Marshall JD, Nethery E, Brauer M. 2008. Within-urban variability in ambient air pollution: Comparison of estimation methods. Atmos. Environ. 42:1359–1369; doi:16/j.atmosenv.2007.08.012. Martinez FD. 2007. CD14, endotoxin, and asthma risk: actions and interactions. Proc. Am. Thorac. Soc. 4:221–225; doi:10.1513/pats.200702-035AW. Martinez FD, Wright AL, Taussig LM, Holberg CJ, Halonen M, Morgan WJ. 1995. Asthma and Wheezing in the First Six Years of Life. N. Engl. J. Med. 332:133–138; doi:10.1056/NEJM199501193320301. Masoli M, Fabian D, Holt S, Beasley R, Global Initiative for Asthma (GINA) Program. 2004. The global burden of asthma: executive summary of the GINA Dissemination Committee report. Allergy 59:469–478; doi:10.1111/j.1398-9995.2004.00526.x. Matsui EC, Eggleston PA, Buckley TJ, Krishnan JA, Breysse PN, Rand CS, et al. 2006. Household mouse allergen exposure and asthma morbidity in inner-city preschool children. Ann. Allergy. Asthma. Immunol. 97:514–520; doi:10.1016/S1081-1206(10)60943-X. McColley SA, Morty RE. 2012. Update in pediatric lung disease 2011. Am. J. Respir. Crit. Care Med. 186:30–34; doi:10.1164/rccm.201203-0568UP. McConnell R, Islam T, Shankardass K, Jerrett M, Lurmann F, Gilliland F, et al. 2010. Childhood Incident Asthma and Traffic-Related Air Pollution at Home and School. Environ Health Perspect 118:1021–1026; doi:10.1289/ehp.0901232. McCurdy LE, Winterbottom KE, Mehta SS, Roberts JR. 2010. Using nature and outdoor activity to improve children’s health. Curr. Probl. Pediatr. Adolesc. Health Care 40:102–117; doi:10.1016/j.cppeds.2010.02.003. Melén E, Nyberg F, Lindgren CM, Berglind N, Zucchelli M, Nordling E, et al. 2008. Interactions between Glutathione S-Transferase P1, Tumor Necrosis Factor, and Traffic-Related Air Pollution for Development of Childhood Allergic Disease. Environ. Health Perspect. 116:1077–1084; doi:10.1289/ehp.11117. Millar WJ, Hill GB. 1998. Childhood asthma. Health Rep. Stat. Can. Can. Cent. Health Inf. Rapp. Sur Santé Stat. Can. Cent. Can. Inf. Sur Santé 10: 9–21 (ENG); 9–22 (FRE). Miller KA, Siscovick DS, Sheppard L, Shepherd K, Sullivan JH, Anderson GL, et al. 2007. Long-term exposure to air pollution and incidence of cardiovascular events in women. N. Engl. J. Med. 356:447–458; doi:10.1056/NEJMoa054409. 172  Mitchell R, Popham F. 2008. Effect of exposure to natural environment on health inequalities: an observational population study. The Lancet 372:1655–1660; doi:10.1016/S0140-6736(08)61689-X. Moffatt MF, Gut IG, Demenais F, Strachan DP, Bouzigon E, Heath S, et al. 2010. A large-scale, consortium-based genomewide association study of asthma. N. Engl. J. Med. 363: 1211–1221. Mölter A, Agius RM, de Vocht F, Lindley S, Gerrard W, Lowe L, et al. 2013. Long-term exposure to PM10 and NO2 in association with lung volume and airway resistance in the MAAS birth cohort. Environ. Health Perspect. 121:1232–1238; doi:10.1289/ehp.1205961. Mölter A, Agius R, Vocht F de, Lindley S, Gerrard W, Custovic A, et al. 2014a. Effects of long-term exposure to PM10 and NO2 on asthma and wheeze in a prospective birth cohort. J. Epidemiol. Community Health 68:21–28; doi:10.1136/jech-2013-202681. Mölter A, Lindley S, de Vocht F, Agius R, Kerry G, Johnson K, et al. 2012a. Performance of a microenviromental model for estimating personal NO2 exposure in children. Atmos. Environ. 51:225–233; doi:10.1016/j.atmosenv.2012.01.030. Mölter A, Lindley S, Vocht F de, Simpson A, Agius R. 2010a. Modelling air pollution for epidemiologic research–Part I: A novel approach combining land use regression and air dispersion. Sci. Total Environ. 408: 5862–5869. Mölter A, Lindley S, Vocht F de, Simpson A, Agius R. 2010b. Modelling air pollution for epidemiologic research - Part II: Predicting temporal variation through land use regression. Sci. Total Environ. 409:211 – 217; doi:DOI: 10.1016/j.scitotenv.2010.10.005. Mölter A, Simpson A, Berdel D, Brunekreef B, Custovic A, Cyrys J, et al. 2014b. A multicentre study of air pollution exposure and childhood asthma prevalence: the ESCAPE project. Eur. Respir. J. erj00836–2014; doi:10.1183/09031936.00083614. Moraes TJ, Lefebvre DL, Chooniedass R, Becker AB, Brook JR, Denburg J, et al. 2015. The Canadian Healthy Infant Longitudinal Development Birth Cohort Study: Biological Samples and Biobanking. Paediatr. Perinat. Epidemiol. 29:84–92; doi:10.1111/ppe.12161. Morales E, Garcia-Esteban R, de la Cruz OA, Basterrechea M, Lertxundi A, de Dicastillo MDML, et al. 2015. Intrauterine and early postnatal exposure to outdoor air pollution and lung function at preschool age. Thorax 70:64–73; doi:10.1136/thoraxjnl-2014-205413. Morgenstern V, Zutavern A, Cyrys J, Brockow I, Gehring U, Koletzko S, et al. 2007. Respiratory health and individual estimated exposure to traffic-related air pollutants in a cohort of young children. Occup. Environ. Med. 64:8 –16; doi:10.1136/oem.2006.028241. Morgenstern V, Zutavern A, Cyrys J, Brockow I, Koletzko S, Krämer U, et al. 2008. Atopic Diseases, Allergic Sensitization, and Exposure to Traffic-related Air Pollution in Children. Am. J. Respir. Crit. Care Med. 177:1331–1337; doi:10.1164/rccm.200701-036OC. 173  Mortimer K, Neugebauer R, Lurmann F, Alcorn S, Balmes J, Tager I. 2008. Early-lifetime exposure to air pollution and allergic sensitization in children with asthma. J. Asthma Off. J. Assoc. Care Asthma 45:874–881; doi:10.1080/02770900802195722. Nagin DS. 2005. Group-Based Modeling of Development. Harvard University Press. Nagin DS, Odgers CL. 2010. Group-Based Trajectory Modeling in Clinical Research. Annu. Rev. Clin. Psychol. 6:109–138; doi:10.1146/annurev.clinpsy.121208.131413. Nethery E, Brauer M, Janssen P. 2008a. Time-activity patterns of pregnant women and changes during the course of pregnancy. J Expos Sci Env. Epidemiol 19: 317–324. Nethery E, Leckie SE, Teschke K, Brauer M. 2008b. From measures to models: an evaluation of air pollution exposure assessment for epidemiological studies of pregnant women. Occup. Environ. Med. 65:579 –586; doi:10.1136/oem.2007.035337. Nicolai T, Carr D, Weiland S, Duhme H, von Ehrenstein O, Wagner C, et al. 2003. Urban traffic and pollutant exposure related to respiratory outcomes and atopy in a large sample of children. Eur. Respir. J. Off. J. Eur. Soc. Clin. Respir. Physiol. 21: 956. Nicolaou N, Siddique N, Custovic A. 2005. Allergic disease in urban and rural populations: increasing prevalence with increasing urbanization. Allergy 60:1357–1360; doi:10.1111/j.1398-9995.2005.00961.x. Nordling E, Berglind N, Melén E, Emenius G, Hallberg J, Nyberg F, et al. 2008. Traffic-related air pollution and childhood respiratory symptoms, function and allergies. Epidemiol. Camb. Mass 19:401–408; doi:10.1097/EDE.0b013e31816a1ce3. Oftedal B, Brunekreef B, Nystad W, Nafstad P. 2007. Residential outdoor air pollution and allergen sensitization in schoolchildren in Oslo, Norway. Clin. Exp. Allergy J. Br. Soc. Allergy Clin. Immunol. 37:1632–1640; doi:10.1111/j.1365-2222.2007.02823.x. Oftedal B, Nystad W, Brunekreef B, Nafstad P. 2009. Long-term traffic-related exposures and asthma onset in schoolchildren in oslo, norway. Environ. Health Perspect. 117:839–844; doi:10.1289/ehp.11491. Olson DA, McDow SR. 2009. Near roadway concentrations of organic source markers. Atmos. Environ. 43:2862–2867; doi:10.1016/j.atmosenv.2009.03.016. Olson DA, Turlington J, Duvall RM, McDow SR, Stevens CD, Williams R. 2008. Indoor and outdoor concentrations of organic and inorganic molecular markers: Source apportionment of PM2.5 using low-volume samples. Atmos. Environ. 42:1742–1751; doi:10.1016/j.atmosenv.2007.11.035. Pakbin P, Ning Z, Schauer JJ, Sioutas C. 2009. Characterization of Particle Bound Organic Carbon from Diesel Vehicles Equipped with Advanced Emission Control Technologies. Environ. Sci. Technol. 43:4679–4686; doi:10.1021/es8030825. 174  Panico L, Stuart B, Bartley M, Kelly Y. 2014. Asthma Trajectories in Early Childhood: Identifying Modifiable Factors. PLoS ONE 9:e111922; doi:10.1371/journal.pone.0111922. Patel MM, Miller RL. 2009. Air pollution and childhood asthma: recent advances  and future directions. Curr. Opin. Pediatr. 21: 235–242. Pattenden S, Hoek G, Braun‐Fahrländer C, Forastiere F, Kosheleva A, Neuberger M, et al. 2006. NO2 and children’s respiratory symptoms in the PATY study. Occup. Environ. Med. 63:828–835; doi:10.1136/oem.2006.025213. Pearce N, Aït-Khaled N, Beasley R, Mallol J, Keil U, Mitchell E, et al. 2007. Worldwide trends in the prevalence of asthma symptoms: phase III of the International Study of Asthma and Allergies in Childhood (ISAAC). Thorax 62: 758–766. Peat JK, Tovey E, Toelle BG, Haby MM, Gray EJ, Mahmic A, et al. 1996. House dust mite allergens. A major risk factor for childhood asthma in Australia. Am. J. Respir. Crit. Care Med. 153: 141–146. Perinatal Services BC [creator]. 2009: BC Perinatal Data  Registry. Perinatal Services BC [publisher]. Data extract. PSBC (2009). http://popdata.bc.ca/data.  Platts-Mills TAE, Erwin EA, Heymann PW, Woodfolk JA. 2009. Pro: The evidence for a causal role of dust mites in asthma. Am. J. Respir. Crit. Care Med. 180:109–113; discussion 120–121; doi:10.1164/rccm.200811-1756PR. Platts-Mills TA, Woodfolk JA, Sporik RB. 2001. Con: the increase in asthma cannot be ascribed to cleanliness. Am. J. Respir. Crit. Care Med. 164: 1107–1108; discussion 1108–1109. Platts-Mills T, Leung DYM, Schatz M. 2007. The role of allergens in asthma. Am. Fam. Physician 76: 675–680. Poplawski K, Gould T, Setton E, Allen R, Su J, Larson T, et al. 2009. Intercity transferability of land use regression models for estimating ambient concentrations of nitrogen dioxide. J. Expo. Sci. Environ. Epidemiol. 19: 107–117. Prosser RJ, Carleton BC, Smith MA. 2008. Identifying persons with treated asthma using administrative data via latent class modelling. Health Serv. Res. 43:733–754; doi:10.1111/j.1475-6773.2007.00775.x. Roemmich JN, Epstein LH, Raja S, Yin L, Robinson J, Winiewicz D. 2006. Association of access to parks and recreational facilities with the physical activity of young children. Prev. Med. 43:437–441; doi:10.1016/j.ypmed.2006.07.007. Rogge WF, Hildemann LM, Mazurek MA, Cass GR, Simoneit BRT. 1993. Sources of fine organic aerosol. 2. Noncatalyst and catalyst-equipped automobiles and heavy-duty diesel trucks. Environ. Sci. Technol. 27:636–651; doi:10.1021/es00041a007. 175  Ronchetti R, Jesenak M, Rennerova Z, Barreto M, Ronchetti F, Villa MP. 2009. Relationship between atopic asthma and the population prevalence rates for asthma or atopy in children: atopic and nonatopic asthma in epidemiology. Allergy Asthma Proc. Off. J. Reg. State Allergy Soc. 30:55–63; doi:10.2500/aap.2009.30.3197. Rook GA. 2013. Regulation of the immune system by biodiversity from the natural environment: An ecosystem service essential to health. Proc. Natl. Acad. Sci. U. S. A. 110:18360–18367; doi:10.1073/pnas.1313731110. Roorda-Knape MC, Janssen NAH, de Hartog J, Van Vliet PHN, Harssema H, Brunekreef B. 1999. Traffic related air pollution in city districts near motorways. Sci. Total Environ. 235:339–341; doi:16/S0048-9697(99)00217-X. Rosa MJ, Jung KH, Perzanowski MS, Kelvin EA, Darling KW, Camann DE, et al. 2011. Prenatal exposure to polycyclic aromatic hydrocarbons, environmental tobacco smoke and asthma. Respir. Med. 105:869–876; doi:10.1016/j.rmed.2010.11.022. Rutter AP, Snyder DC, Schauer JJ, Sheesley RJ, Olson MR, DeMinter J. 2011. Contributions of resuspended soil and road dust to organic carbon in fine particulate matter in the Midwestern US. Atmos. Environ. 45:514–518; doi:16/j.atmosenv.2010.10.014. Ryan PH, LeMasters G, Biagini J, Bernstein D, Grinshpun SA, Shukla R, et al. 2005. Is it traffic type, volume, or distance? Wheezing in infants living near truck and bus traffic. J. Allergy Clin. Immunol. 116:279–284; doi:10.1016/j.jaci.2005.05.014. Ryan PH, LeMasters GK. 2007. A review of land-use regression models for characterizing intraurban air pollution exposure. Inhal. Toxicol. 19 Suppl 1:127–133; doi:10.1080/08958370701495998. Ryan PH, Lemasters GK, Biswas P, Levin L, Hu S, Lindsey M, et al. 2007. A comparison of proximity and land use regression traffic exposure models and wheezing in infants. Environ. Health Perspect. 115:278–284; doi:10.1289/ehp.9480. Salam MT, Li Y-F, Langholz B, Gilliland FD. 2004. Early-life environmental risk factors for asthma: findings from the Children’s Health Study. Environ. Health Perspect. 112: 760–765. Sarnat SE, Coull BA, Ruiz PA, Koutrakis P, Suh HH. 2006. The Influences of Ambient Particle Composition and Size on Particle Infiltration in Los Angeles, CA, Residences. J. Air Waste Manag. Assoc. 56:186–196; doi:10.1080/10473289.2006.10464449. Savenije OE, Granell R, Caudri D, Koppelman GH, Smit HA, Wijga A, et al. 2011. Comparison of childhood wheezing phenotypes in 2 birth cohorts: ALSPAC and PIAMA. J. Allergy Clin. Immunol. 127:1505–1512.e14; doi:10.1016/j.jaci.2011.02.002. Schauer JJ. 2003. Evaluation of elemental carbon as a marker for diesel particulate matter. J. Expo. Anal. Environ. Epidemiol. 13:443–453; doi:10.1038/sj.jea.7500298. 176  Schram-Bijkerk D, Doekes G, Boeve M, Douwes J, Riedler J, Ublagger E, et al. 2006. Exposure to microbial components and allergens in population studies: a comparison of two house dust collection methods applied by participants and fieldworkers. Indoor Air 16: 414–425. Schwartz J, Slater D, Larson TV, Pierson WE, Koenig JQ. 1993. Particulate air pollution and hospital emergency room visits for asthma in Seattle. Am. Rev. Respir. Dis. 147:826–831; doi:10.1164/ajrccm/147.4.826. Sears MR, Greene JM, Willan AR, Wiecek EM, Taylor DR, Flannery EM, et al. 2003. A Longitudinal, Population-Based, Cohort Study of Childhood Asthma Followed to Adulthood. N. Engl. J. Med. 349:1414–1422; doi:doi: 10.1056/NEJMoa022363. Selevan SG, Kimmel CA, Mendola P. 2000. Identifying critical windows of exposure for children’s health. Environ. Health Perspect. 108 Suppl 3: 451–455. Setton E, Marshall JD, Brauer M, Lundquist KR, Hystad P, Keller P, et al. 2010. The impact of daily mobility on exposure to traffic-related air pollution and health effect estimates. J. Expo. Sci. Environ. Epidemiol. Shea KM, Truckner RT, Weber RW, Peden DB. 2008. Climate change and allergic disease. J. Allergy Clin. Immunol. 122:443–453; quiz 454–455; doi:10.1016/j.jaci.2008.06.032. Sheesley RJ, Schauer JJ, Garshick E, Laden F, Smith TJ, Blicharz AP, et al. 2009. Tracking personal exposure to particulate diesel exhaust in a diesel freight terminal using organic tracer analysis. J. Expo. Sci. Environ. Epidemiol. 19:172–186; doi:10.1038/jes.2008.11. Sheppard L, Burnett RT, Szpiro AA, Kim S-Y, Michael Jerrett, C Arden Pope III, et al. 2011. Confounding and exposure measurement error in air pollution epidemiology. Air Qual. Atmosphere Health 5:203–216; doi:10.1007/s11869-011-0140-9. Sherriff A, Maitra A, Ness AR, Mattocks C, Riddoch C, Reilly JJ, et al. 2009. Association of duration of television viewing in early childhood with the subsequent development of asthma. Thorax 64:321–325; doi:10.1136/thx.2008.104406. Sidenius KE, Hallas TE, Brygge T, Poulsen LK, Mosbech H. 2002. House dust mites and their allergens at selected locations in the homes of house dust mite-allergic patients. Clin. Exp. Allergy J. Br. Soc. Allergy Clin. Immunol. 32: 1299–1304. Simoneit BRT. 1985. Application of Molecular Marker Analysis to Vehicular Exhaust for Source Reconciliations. Int. J. Environ. Anal. Chem. 22: 203. Simpson A, Tan VYF, Winn J, Svensén M, Bishop CM, Heckerman DE, et al. 2010. Beyond Atopy. Am. J. Respir. Crit. Care Med. 181:1200–1206; doi:10.1164/rccm.200907-1101OC. 177  Sly PD, Boner AL, Björksten B, Bush A, Custovic A, Eigenmann PA, et al. 2008. Early identification of atopy in the prediction of persistent asthma in children. The Lancet 372:1100–1106; doi:10.1016/S0140-6736(08)61451-8. Sporik R, Holgate ST, Platts-Mills TA, Cogswell JJ. 1990. Exposure to house-dust mite allergen (Der p I) and the development of asthma in childhood. A prospective study. N. Engl. J. Med. 323:502–507; doi:10.1056/NEJM199008233230802. Spycher BD, Silverman M, Brooke AM, Minder CE, Kuehni CE. 2008. Distinguishing phenotypes of childhood wheeze and cough using latent class analysis. Eur. Respir. J. 31:974–981; doi:10.1183/09031936.00153507. Statistics Canada. 2001. Census of Canada 2001 Archive. Stein RT, Martinez FD. 2004. Asthma phenotypes in childhood: lessons from an epidemiological approach. Paediatr. Respir. Rev. 5:155–161; doi:10.1016/j.prrv.2004.01.007. Strachan DP. 2000. Family size, infection and atopy: the first decade of the “hygiene hypothesis.” Thorax 55: S2–S10. Strickland MJ, Klein M, Flanders WD, Chang HH, Mulholland JA, Tolbert PE, et al. 2014. Modification of the Effect of Ambient Air Pollution on Pediatric Asthma Emergency Visits: Susceptible Subpopulations. Epidemiology 25:843–850; doi:10.1097/EDE.0000000000000170. Subbarao P, Becker A, Brook JR, Daley D, Mandhane PJ, Miller GE, et al. 2009. Epidemiology of asthma: risk factors for development. Expert Rev. Clin. Immunol. 5:77–95; doi:10.1586/1744666X.5.1.77. Su JG, Brauer M, Ainslie B, Steyn D, Larson T, Buzzelli M. 2008. An innovative land use regression model incorporating meteorology for exposure analysis. Sci. Total Environ. 390:520–529; doi:16/j.scitotenv.2007.10.032. Sunyer J, Spix C, Quénel P, Ponce-de-León A, Pönka A, Barumandzadeh T, et al. 1997. Urban air pollution and emergency admissions for asthma in four European cities: the APHEA Project. Thorax 52: 760–765. Takaro TT, Scott JA, Allen RW, Anand SS, Becker AB, Befus DA, et al. 2015. The Canadian Healthy Infant Longitudinal Development(CHILD) birth cohort study: assessment of environmental exposures. J Expo Sci Env. Epidemiol; doi:doi: 10.1038/jes.2015.7. Talukder S, Capon A, Nath D, Kolb A, Jahan S, Boufford J. 2015. Urban health in the post-2015 agenda. The Lancet 385:769; doi:10.1016/S0140-6736(15)60428-7. Thurston G, Lippmann M, Scott M, Fine J. 1997. Summertime haze air pollution and children with asthma. Am J Respir Crit Care Med 155:654–660. 178  Torrent M, Sunyer J, Muñoz L, Cullinan P, Iturriaga MV, Figueroa C, et al. 2006. Early-life domestic aeroallergen exposure and IgE sensitization at age 4 years. J. Allergy Clin. Immunol. 118:742–748; doi:10.1016/j.jaci.2006.04.059. To T, Dell S, Dick PT, Cicutto L, Harris JK, MacLusky IB, et al. 2006. Case verification of children with asthma in Ontario. Pediatr. Allergy Immunol. 17:69–76; doi:10.1111/j.1399-3038.2005.00346.x. To T, Gershon A, Wang C, Dell S, Cicutto L. 2007. Persistence and remission in childhood asthma: A population-based asthma birth cohort study. Arch. Pediatr. Adolesc. Med. 161:1197–1204; doi:10.1001/archpedi.161.12.1197. Tovey ER, Almqvist C, Li Q, Crisafulli D, Marks GB. 2008. Nonlinear relationship of mite allergen exposure to mite sensitization and asthma in a birth cohort. J. Allergy Clin. Immunol. 122:114–118, 118.e1–5; doi:10.1016/j.jaci.2008.05.010. Turlington JM, Olson DA, Stockburger L, McDow SR. 2010. Trueness, precision, and detectability for sampling and analysis of organic species in airborne particulate matter. Anal. Bioanal. Chem. 397:2451–2463; doi:10.1007/s00216-010-3833-2. Tzivian L. 2011. Outdoor Air Pollution and Asthma in Children. J. Asthma Off. J. Assoc. Care Asthma; doi:10.3109/02770903.2011.570407. Venn AJ, Lewis SA, Cooper M, Hubbard R, Britton J. 2001. Living near a main road and the risk of wheezing illness in children. Am. J. Respir. Crit. Care Med. 164: 2177–2180. Vineis P, Hoek G, Krzyzanowski M, Vigna‐Taglianti F, Veglia F, Airoldi L, et al. 2006. Air pollution and risk of lung cancer in a prospective study in Europe, Air pollution and risk of lung cancer in a prospective study in Europe. Int. J. Cancer Int. J. Cancer 119, 119:169, 169–174, 174; doi:10.1002/ijc.21801, 10.1002/ijc.21801. Von Hertzen L, Haahtela T. 2009. Con: House dust mites in atopic diseases: accused for 45 years but not guilty? Am. J. Respir. Crit. Care Med. 180:113–119; discussion 119–120; doi:10.1164/rccm.200807-1001CO. Von Mutius E. 2009. Gene-environment interactions in asthma. J. Allergy Clin. Immunol. 123:3–11; quiz 12–13; doi:10.1016/j.jaci.2008.10.046. Von Mutius E. 2001. Pro: the increase in asthma can be ascribed to cleanliness. Am. J. Respir. Crit. Care Med. 164: 1106–1107; discussion 1108–1109. Von Mutius E, Martinez F, Fritzsch C, Nicolai T, Roell G, Thiemann H. 1994. Prevalence of asthma and atopy in two areas of West and East Germany. Am J Respir Crit Care Med 149:358–364. Vork KL, Broadwin RL, Blaisdell RJ. 2007. Developing asthma in childhood from exposure to secondhand tobacco smoke: insights from a meta-regression. Environ. Health Perspect. 115: 1394–1400. 179  Wallace L, Williams R. 2005. Use of personal-indoor-outdoor sulfur concentrations to estimate the infiltration factor and outdoor exposure factor for individual homes and persons. Environ. Sci. Technol. 39: 1707–1714. Wang R, Henderson SB, Sbihi H, Allen RW, Brauer M. 2013. Temporal stability of land use regression models for traffic-related air pollution. Atmos. Environ. 64:312–319; doi:10.1016/j.atmosenv.2012.09.056. Weier J, Herring D. 2000. NASA Earth Observatory : Measuring Vegetation (NDVI & EDI). Available: http://earthobservatory.nasa.gov/Features/MeasuringVegetation/measuring_vegetation_1.php [accessed 22 October 2014]. Weinmayr G, Weiland SK, Björkstén B, Brunekreef B, Büchele G, Cookson WOC, et al. 2007. Atopic Sensitization and the International Variation of Asthma Symptom Prevalence in Children. Am. J. Respir. Crit. Care Med. 176:565–574; doi:10.1164/rccm.200607-994OC. Wheeler AJ, Smith-Doiron M, Xu X, Gilbert NL, Brook JR. 2008. Intra-urban variability of air pollution in Windsor, Ontario--Measurement and modeling for human exposure assessment. Environ. Res. 106:7–16; doi:16/j.envres.2007.09.004. Wheeler AJ, Xu X, Kulka R, You H, Wallace L, Mallach G, et al. 2011. Windsor, Ontario exposure assessment study: design and methods validation of personal, indoor, and outdoor air pollution monitoring. J. Air Waste Manag. Assoc. 1995 61: 324–338. Whitehead T, Metayer C, Gunier RB, Ward MH, Nishioka MG, Buffler P, et al. 2011. Determinants of polycyclic aromatic hydrocarbon levels in house dust. J. Expo. Sci. Environ. Epidemiol. 21:123–132; doi:10.1038/jes.2009.68. Whitehead TP, Metayer C, Petreas M, Does M, Buffler PA, Rappaport SM. 2013. Polycyclic Aromatic Hydrocarbons in Residential Dust: Sources of Variability. Environ. Health Perspect. 121:543–550; doi:10.1289/ehp.1205821. Whitehead TP, Nuckols JR, Ward MH, Rappaport SM. 2012. Carpet-dust chemicals as measures of exposure: Implications of variability. Emerg. Themes Epidemiol. 9:2; doi:10.1186/1742-7622-9-2. Wong EY, Gohlke J, Griffith WC, Farrow S, Faustman EM. 2004. Assessing the health benefits of air pollution reduction for children. Environ. Health Perspect. 112: 226–232. Wong GWK, Chow CM. 2008. Childhood asthma epidemiology: Insights from comparative studies of rural and urban populations. Pediatr. Pulmonol. 43:107–116; doi:10.1002/ppul.20755. World Health Organization Regional Office for Europe. 2013. Review of evidence on health aspects of air pollution – REVIHAAP project: final technical report. 180  Wu H, Romieu I, Sienra-Monge J-J, del Rio-Navarro BE, Anderson DM, Dunn EW, et al. 2007. Parental Smoking Modifies the Relation between Genetic Variation in Tumor Necrosis Factor-α (TNF) and Childhood Asthma. Environ. Health Perspect. 115:616–622; doi:10.1289/ehp.9740. Wu P, Carroll KN, Gebretsadik T, Dupont WD, Mitchel EF, Hartert TV. 2012. The developmental trajectory of pediatric asthma in 3- to-10-year-olds. J. Allergy Clin. Immunol. 129:1397–1398; doi:10.1016/j.jaci.2012.01.029. Zmirou D, Gauvin S, Pin I, Momas I, Sahraoui F, Just J, et al. 2004. Traffic related air pollution and incidence of childhood asthma: results of the Vesta case-control study. J. Epidemiol. Community Health 58:18 –23; doi:10.1136/jech.58.1.18.    181  Appendices19  Appendix A   Allergy skin prick testing  A.1 Performing the allergy skin prick test A.1.1 Allergy skin prick supplies • Allergens (ALK Abello) • Duotip II Dipwell Trays (Lincoln Diagnostics) • Duotips II (Lincoln Diagnostics) • Bic Ultra Round Stic Grip pen for circling wheals • 3M Scotch® Tear-by-hand™ Tape 105-NA, 48 mm  x 16 mm • Antiseptic Isopropyl Alcohol Pad • Tissues • Epipen • Stop watch or timer • Allergy Skin Test Record Forms – Skin Test Worksheet A.1.2 Quality control of allergens  Some allergens may not expire for up to 3 years. The variability of allergen effectiveness at room, fridge and Dipwell temperature is unknown. For this reason we will replace each antigen yearly at all centers. The central site will order the same lot numbers for all study centers on a yearly basis and keep logs to track lot numbers and expiry dates                                                      19 CHILD Standard Operating Procedure from © CHILD Version date May 15 2010; accessed from health diary resources, adapted with permission from Principal Investigator.  182  A.1.3 Timeline and checklist    DAD  MOM  CHILD    Allergen (ALK Item #)   18 wk    1 year  2 year    3 year    5 year 1 year (PFT) 1. Histamine       2. Glycerin Control       3. Alternaria alternate       4. Cladosporium       5. Penicillium Mixed       6. Aspergillus fumigatus       7. Cat Hair, Standardized       8. Dog Epithelium       9. D. Pteronyssinus       10. D. Farinae       11. Cockroach, German       12. Trees Midwest       13. Grass Mix       14. Weeds       15. Ragweed Mixed       16. Peanut       17. Milk, Whole Cow’s       18. Egg White       19. Soybean        Child-Dipwell Tray 2: Labels at 1 year   9. D. Pteronyssinus 8. Dog Epithelium 7. Cat Hair, Standardized 3.Alternaria alternata 2.Glycerin Control   1.    19. Soybean 18. Egg White 17. Milk, Whole Cow’s  16. Peanut 11. Cockroach, 10. D. Farinae  183  A.1.4 Skin testing procedure Directions for using the Dipwell Tray II: The tray consists of 40 clear plastic wells that hold test solutions and Duotip-Test II devices. The 40 wells are arranged in four rows of ten, with each row identified by letters A through D. i. Type or print the appropriate extract or control information in the blank boxes provided on the label. Apply labels over designated areas (each label contains 10 rectangular spaces). ii. Remove sterile wells from their package of 40 and insert 16 wells (2 rows of 8). Use enough pressure when placing wells into the receiving holes to push them downward as far as they will travel. iii. Individual test solutions are placed into individual wells so that each well contains the test solution that corresponds with that test solution’s well location shown on the label. iv. Dispense 0.25 ml of each skin test solution into its properly identified and numbered well, using the dropper in the 10 mL vial. The dropper must be inserted down into the well. Each well must contain enough volume to allow complete submersion of the Duotip II (at least 4 drops of solution from the vial dropper). v. Place clean duotips in the wells and place the lid securely over duotips/ wells and keep refrigerated when not in use. Remove tray with wells 15 minutes before skin testing and keep at room temperature.  184   A.1.4.1 Procedure for skin prick test 1. Explain procedure to child’s mother or father as appropriate. Provide a Skin Prick Test Result Informational Handout (Appendix A.2 Figure 2) 2. Ascertain whether the child has taken medications or herbal supplements in the last 7 days. Give medication/supplement prompt list (Appendix A.2 Supplementary Material 2) as a prompt and fill out the medication/supplement skin test form (Appendix A.2 Figure 3). 3. Seat the child in the parent’s lap if possible and place their forearm in a supine position on a flat surface. 4. Clean and dry inner aspect of child’s forearms with an alcohol swab (70% isopropyl alcohol). 5. Make small dots with pen on the radial aspect of the forearm, in rows corresponding to the numbers of allergens that will be used. Include the controls in the total number. Use a folded photocopied allergy skin test sheet to space the dots by placing the dot directly beside the numbers.    6. Identify the allergens by corresponding allergen numbers on the radial aspect of the 185  appropriate dot so that the labelling will not be included in the results. 7. Place one sterile Duotip II in each allergen well. Follow the order on the skin test form 8. At a site 2 cm medial (ulnar) to the pen mark, using a sterile Duotip for each allergen, hold device so its shaft forms a 45 degree angle with the skin plane. Prick with one point and lift skin simultaneously. Do not prick twice. 9. Discard Duotip in the appropriate sharps container. 10. Blot (DO NOT WIPE) skin dry with tissue making sure not to cross contaminate allergens. 11. Instruct child not to scratch the area and to remain in the waiting area until tests are read. 12. Ten minutes after the first (Histamine) prick, use a Bic pen to outline the wheal of the Histamine. A wheal response, also called a welt, is a firm elevated swelling of the skin. Use your fingers to feel the swelling or bump to help identify a wheal response. After 20 minutes, outline wheal responses for the allergens. Wheal tracings must be closed without any gaps. Do not record the flare. If there is no wheal, do not make any marks. Use cross-lighting if needed to identify the edges of the wheals.  13. Using transparent adhesive tape, cut a piece large enough from the roll to cover the histamine outline, press gently, and remove. 14. Apply tape to the skin test form. 15. Repeat for all other allergy wheals that are outlined. 16. All allergy sheets are bar coded and numbered specifically for each case. Tape must adhere completely flat (ideally without any creases and without hair or other material attached). 186  Wheal tracings cannot cross the allergen box or touch the yellow box on the skin test form. (See Appendix A.2 Figure 5, for an example of a correctly completed form). 17. If there is hair attached to the tape that you cannot readily remove: a. Stick the tape to a BLANK piece of paper. b. Place another piece of tape, sticky side up, and trace the outline on the sticky side of the new piece of tape. c. Transfer the new piece of tape with copied outline to the skin test worksheet. 18. Wipe the arm with an alcohol swab to remove ink and to help decrease itch.  Allergy Skin Testing for the child at 1 year should not be done in the field. The child’s test must be done in the clinic A.1.4.2 Taking care of allergens and Epipen instructions Testing will be to common outdoor allergens, indoor allergens and food allergens. The testing will be carried out by the study personnel listed as per SOP. The local site investigator or his/her physician delegate must be aware of the testing and be available if an adverse event occurs. If the investigator is not available, an alternate contact approved by the investigator must be available. FALSE POSITIVES may occur if… i. the prick is too hard ii. patient has sensitive skin (i.e. Dematographism) iii. there is cross contamination between allergens  FALSE NEGATIVES may occur if… i. the antigen site is not pricked hard enough or skin is not lifted ii. antihistamines or medication with antihistamine activity (e.g., Gravol, antidepressants such 187  as amitryptiline) have not been withheld for the appropriate length of time  CARING FOR ALLERGENS i. Refrigerate allergens when not in use. ii. Take  the  allergens  out  of  the  fridge  15  minutes  before  using  and  leave  at  room temperature. iii. Take great care to return dropper into correct vial to avoid cross contamination. iv. Discard allergen if the extracts look cloudy.  SAFETY PRECAUTIONS: Always have two Epipens or Twinjects available at the testing site and check the expiry dates regularly. If systemic symptoms including urticaria, rhinitis, conjunctivitis, angioedema, cough, wheezing, faintness or hypotension occur, give 1 injection of Epipen intramuscularly. Obtain vital signs. Ensure that the physician attends immediately. Repeat the dose of Epipen after 5 or 10 minutes if necessary. An Epipen Jr is a disposable, prefilled automatic injection device used for allergic emergencies. They contain a single dose of epinephrine which is injected into the outer thigh. It has an expiry date and should be replaced before this date. It should be kept at room temperature. Do not store in refrigerator. For complete instructions see Supplementary material 1: Epipen instructions.  A.1.4.3 Measurements of Wheals 1. Introduction and Purpose: To measure the taped skin prick reactions to assess individual hypersensitivity to allergens. 2. Scope: This SOP applies to the activities involved in assessing children enrolled in the CHILD Study. 3. Responsibility: The Data Manager at the Winnipeg site will ensure appropriate final analysis of allergy skin test measurements using computer analysis and recognition. 4. Measurement Procedure 188  a. The allergy forms are bar coded and numbered specifically for each study participant. b. Skin tests will be initially measured in Hamilton and recorded on the subject’s Skin test data form (Figure 4, page 23). c. Measurements will be taken of the wheal only and be made to the nearest 0.5 mm. d. The length (1st line, seen below) of the wheal is measured first, followed by the width measured at the midpoint of, and perpendicular to, the length (2nd line, seen  below). e. The length and width will be double entered into HealthDiary. f. Original forms will then be sent to the Winnipeg Pediatric Allergy Lab, and scanned into the Winnipeg database.  Examples (enlarged): 1st line    2nd line   189   A.2 Figures and supplemental materials A.2.1 Appointment Letter – Child  Dear Ms.    Thank  you  for  your  willingness  to  participate  in  the  Canadian  Healthy  Infant  Longitudinal Development (CHILD) Study. We look forward to meeting you at the clinic. We have arranged an appointment for your child to have a clinical assessment and an allergy skin test and to give a sample of blood. We will see you at PLACE DAY  DATE  TIME If you need to change this appointment, please call the number at the end of the letter Please read the following information regarding the allergy and breathing tests. What is an allergy skin prick test? The test is done to see if you have sensitivities to common substances such as molds, pollens, animal dander, dust mites and some foods to name a few. Before the test, you will be asked about your child’s medication history and any antihistamines he/she may have taken. A series of 12 pricks will be made on the child’s arm with plastic Duotip devices. Each one carries a drop of an allergen into the superficial layer of the skin. You are asked to wait 20 minutes and then we measure any reactions on your child’s arm. A reaction looks very similar to a mosquito bite, with itchiness, redness and swelling of the skin. What is involved in the clinical assessment? The study (paediatrician, team, nurse, staff etc ) will examine your child’s skin for any signs of skin redness or inflammation, listen to his/her chest and take a height, weight and head circumference measurement. HOW SHOULD I PREPARE MY CHILD FOR THE APPOINTMENT? The study staff would also like to collect a stool sample, urine sample and nasal swab. For 4 days before the appointment: 190  Please do not give your child any anti-histamines (medications such as Hismanol, Benadryl, cough medications, Gravol or motion sickness medication) as these will affect the allergy skin prick test results. When you come in to the clinic, bring all your child’s medications with you including any over the counter medication, as treatments may affect the results of these tests in several ways, and so we need to know exactly what your child is taking. I have attached a map, directions and our contact numbers.  We look forward to seeing you. Sincerely;  Dr. [site coordinator] For any change of appointment, call    191   A.2.2 Informational Skin Test Handout192   A.2.3 Child medications and supplements form  193  A.2.4 Skin test measurement form 194   A.2.5 Skin test result form  195   A.2.6 Supplementary material 1: Epipen instructions 196   A.2.7 Supplementary material 2: medications and supplements Prompt List Medications and Supplements that affect skin test. 197   Appendix B : Home assessment (including dust sampling) B.1 Preparation of the visit B.1.1 Goals  The research  coordinator  will  call  the  mother  to  set  up  a  3-month  home  visit 2 weeks in advance and will follow the phone script template. Ideally, the research coordinator who recruited the family into the study will make the first postnatal telephone call approximately 1-2 months after delivery. The goals are: • To reconnect with the family. • To ensure that the family is still eligible (i.e. has not moved, has not refused to further participate, etc.). • To arrange for a home visit: to discuss possible dates and times that might be feasible for a visit, and to confirm the address. • To go over home visit package instructions for breast milk collection and stool collection • To ask the mother’s to fill out the Environmental home questionnaire • To remind the parent about collection of house dust from the bed the child sleeps in and the most often used living area The participant is reminded about the confidentiality and all measures taken to maintain their information non-identifiable. Two days before the visit, phone to confirm the visit and ask the mother if she has her breast milk container. Remind the mother to start collecting breast milk a day before the visit, and to save a soiled diaper and a wet diaper from the day before the visit.  198  B.1.2 Before the home visit B.1.2.1 Home assessment kit protocol 1. Prepare Home Assessment Kit Sanitaire Canister Vac: Model S3680 (Serial#074002221) a) Connection Hose b) Telescoping Wand c) Vacuum Bag d) Vacuum Head - depyrogenated e) Sterile Tweezers f) Sterile scissors 2. Cooler 3. Ice Pack (not included) 4. Tool Bag a) Laser Distance Meter b) GPS c) Flashlight d) Tape Measure e) Green Painter’s Tape f) Scissors g) Pens (not included) h) Sharpie i) Extra Batteries 5. Two Clipboards “Hognose” © Vacuum Head 199  6. Driver’s Log Book 7. Dirty Equipment Bag 8. Extra Soiled Diaper Bag 9. Sample Collection Box a) Refractometer b) 20 mL Syringe c) 3 Cotton Pads d) Nitrile Gloves e) Respirator Mask (3M 8210 N95) f) Alcohol Wipes 10. Map book (not included) 11. Samples’  Bag a) 6 pre-labelled cryovials: See SOP b) Extra Breast Milk Bottle (x1): See SOP c) Dust Bottles (x4) d) 4 stool pre-labelled cryovials: See SOP e) Nasal Swab and Media (x1) and 6 pre-labelled cryovials: See SOP 12. Extra Biohazard Bags 13. Postnatal Questionnaire Package B.1.2.2  Protocol for cleaning  - Vacuum heads and tweezers before visit: • Rinse the dirty vacuum head and tweezers with water for 5 minutes.  • Remove the Teflon wheels of the vacuum head • Wrap both the head (without wheels) and tweezers together in tin foil. 200  • Put both in the laboratory oven (Fisher Oven ISOTMP STD 2.5 Cubic ft. Cat @ 13 247 625G) and bake at 220°C for 2 hours. Remove from oven and allow to cool. - Protocol for cleaning vacuum: After use, clean the vacuum, especially the wheels, with an alcohol wipe to prevent dust contamination from house to house. - Protocol for cleaning refractometer:  • After use, rinse the refractometer with water.  • Pat dry with a paper towel.  • Wipe with an alcohol swab. B.1.2.3  HealthDiary  Print out sample collection forms and labels from HealthDiary. These forms will contain pre-printed information including the subject’s Identification number (ID), the visit (3MON), the sample type (NASAL, URINE, STOOL, BRSMILK, MUDUST, CBDUST), and a summary of collection instructions (see Appendix B.3 Figure 1, for example of completed Dust Collection Form). • 6 aliquot labels for urine cyrovials • 1 labeled for extra breast milk bottle (One is given to mother in advance at the birth visit) • 1 sample (collection) label for diaper bag, 4 cryrovial (aliquot) labels • 1 label for nasal sample vial, and 6 labels for cryovials.  • 2 sample labels for the dust collection bottles with thimbles for the Most Used Living Area (MUDUST). • 2 sample labels for the dust sample collection bottles with thimbles for the child’s sleeping area and bed (CBDUST; see Appendix B.2 Figure 2). Ensure ice packs are frozen for the visit. Take cooler with ice packs in them to the visit. 201  B.1.3 At the home visit Administer questionnaires and conduct home assessment:  3-Months Questionnaires/ Task  HD Code  Version Q:  Residences to 3 Months Q160RESIDEN3M 2009M11D11 Q: Home Environment 3 Months Q114HENVQ3M 2009M09D15 Q: Environmental Assessment RA 3 Months Q117ENVRA3M 2009M08D07 Q: Mother's Nutrition  -- Q: Mother's Meds-Birth to 3 Months Q151MMED3M 2009M08D31 Q: Mother's Vitamins and Supplements 3 Months  Q154MSUPPL3M  2009M11D30 Q: Child Health 3 Months Q127CHLTH3M 2009M09D15 Q: Child Medications-Birth to 3 Months Q148CMED3M 2009M08D31 TCF: Mother Breast Milk 3 Months BRSTMLK- F55SPECIMEN 2008M12D15 TCF: Child Nasal 3 Months Nasal-F37SPECIMEN 2008M12D15 TCF: Child Urine 3 Months Urine-F61SPECIMEN 2008M12D15 TCF: Child Stool 3 Months Stool-F37SPECIMEN 2008M12D15 TCF: Dust Child's Bedroom 3 Months CBDust-F47SPECIMEN 2008M12D15 TCF: Dust Most-Used Living Space 3 Months MUDust- F47SPECIMEN  2008M12D15 TCF: Control Dust 3 Months Urine QC- F61SPECIMEN  2008M12D15   B.1.3.1 Dust samples • Two thimbles of dust to be collected from the most used living area and two thimbles  to be collected from the child’s sleeping area (bed and floor). Two dust samples in total will be collected (from three sites): The first sample to be taken is a combined dust sample collected from the bed that baby sleeps (could be mom’s bed if the baby sleeps with her) and the floor in the room where baby sleeps. A separate sample will be collected from the most used living area room floor. • In conjunction with the home inspection, a dust sample will be collected from a 2- metre square area of floor (or greater area if necessary) using the above standardized consumer model 202  vacuum cleaner fitted with a dust collection device. This area is created by four separate squares of 0.71 m x 0.71 m. A mask may be worn if sensitive to dust. If there is not enough carpet, vacuum available carpet and check thimbles to see if full. If the thimble is full, no more vacuuming is necessary. Record the exact area in HealthDiary. If there is not enough dust, continue vacuuming any other flooring type and record the exact area. • If the flooring is hardwood or laminate, the research assistant will vacuum as much area as possible with effort made to collect dust from the corners/edges of the room. Carpeted area > 2 square metres total  1. Tape off 4 square areas of 0.71 m x 0.71 m in separate locations.  (not in direct sunlight) Hardwood Only  1.Vacuum whole floor room to fill dust thimbles.  Measure and record the exact area vacuumed in HealthDiary. Area rug < 2 square metres  1. If there is an area rug less than 2 square metres, vacuum rug, check thimble. If there is enough dust, document the exact area vacuumed.  2. If there is not enough dust, vacuum hardwood area of the room and document total exact area of rug and  First dust sample will be taken from the child’s bedding and bedroom floor:  • Don a pair of nitrile gloves. Open the pyrogen-free sampling package. Remove the bag containing the tweezers, and carefully open the bag at the closed end of the tweezers. Do not touch the open end of the tweezers. Carefully remove the clean sampler head. Using the tweezers, place two collection thimbles into the sampler head, taking care to ensure that the parts are touched only by the tweezers. Connect the sampler head to the vacuum. Place the tweezers back into the sterile bag/tinfoil, with the open end of the tweezers going in first, and set them aside in an undisturbed place. • Plug the vacuum cleaner into the nearest electrical outlet, using the extension cord if necessary. Try not to unplug household items, especially if it is something that requires being reset, such as a clock, microwave, or VCR. If you must unplug something, try to unplug a lamp. Do not 203  unplug refrigerators or freezers unless you are immediately able to plug it into a power strip. Be careful not to walk or drag the cord or vacuum through the floor area to be vacuumed (floor template area). • Pull back the bed covers and the top sheet, and vacuum the bottom sheet. • If the child sleeps in the parents’ bed, vacuum the half of the bed usually used by the child. If a pillow is used, vacuum the pillow with the case on (the outside of the case), then remove case and vacuum the pillow or the subsequent cover (for a total 1.5-2 minutes; use timer). If child sleeps in a bassinette with a plastic- covered mattress, vacuum the cover or sheet. • Using the same vacuum head and collection thimbles, the floor sample will be taken from four 0.71-metre squares that are each at different distances from the door of the room. This method is used when floor is carpeted. • If there is not enough floor space in the child‟s bedroom, add the room in which the child spends the next most amount of time. • With tape, mark four 0.71-metre squares for the sample. • Choose the locations for the four sampling squares in the bedroom as follows: o One square on any cleared area where the child is likely to play. o Three other squares that are away from direct sunlight, outside the main foot-traffic pattern, and at different distances from door (this will often not be possible). • Vacuum in one direction in a straight line inside the template, making 10 passes back-and-forth along each line until the area has been completely vacuumed. Using the same sampling filter, move to the next square and sample it in the same way until all four squares have been sampled. Do not include the area directly under the bed or wet areas. • Holding the vacuum head with the screws facing up, unscrew the plate, remove the thimble using the sterile tweezers, and estimate the mass of dust collected. If the thimble is not full, reassemble 204  and continue to vacuum in the prescribed way, adding 0.71-metre squares as needed to obtain as much dust as is required. For some hard or very clean floors, you will need to vacuum the entire floor area (e.g. hardwood or laminate flooring). Measure and record the area vacuumed in this circumstance. • When dust collection is complete, carefully open the labelled storage vial for the bedroom sample without touching the inside of the lid. • Holding the vacuum head with the screws facing up, unscrew the plate, remove the thimble using the sterile tweezers, and transfer the thimble into the vial. Close the vial and store it in the transport bag. Second sample from the most used living area in which child plays on the floor Lay out four 0.71-metre squares with tape. If an area rug is present, use this first to reach the goal of a 2-metre square. Vacuum in the same manner as in the bedroom. Observe the thimbles for dust quantity. Once 2 m2 is vacuumed, hold the vacuum head with the screws facing up, unscrew the plate, remove the thimbles using the sterile tweezers, and transfer the thimbles into a storage vial labelled “most used living area” (MULA). Leave the room as you found it. After collecting all samples required in a room, reconnect any lamps or other electrical devices that were disconnected. Remake bed as you found it, replacing pillows or stuffed animals in the same location. If the bed was unmade, leave it unmade. Replace any items you moved from the bed or floor surface in order to take the dust samples. B.1.3.2  Volume measurements Use a work sheet to figure out the volume of the home (Figure 3: worksheet example).  Do not measure individual closets Do not measure every room or hall way. For basement volumes, if possible foundation length x width x height 205  B.1.3.3 Take GPS measurements 1) Record position while standing at the front door. If a signal cannot be found at the front door, attempt to acquire coordinates at a point nearest to the front door on the perimeter of the residence.  2) Turn on the GPS using the ‘power’ button on the bottom right.  3) Once the ‘tracking satellites’ screen comes on (takes approx 5 seconds) push the top right button, labeled “page” 4 times.  The Menu screen will come up and ‘Mark’ will be highlighted  4) Push the ‘enter’ button (bottom left) and the coordinates will be displayed.  5) Record coordinates on questionnaire.  6) Record accuracy measurement 7) Ensuring proper set up of the GPS The reading should be shown as: N   °  ’   . ” W               °    __’         .    ” (See picture above) If the reading is NOT shown as above, please correct format by the following: 8) Go back to menu by pressing the page button, top right. 9) Select “setup” on menu by using top left scroll buttons 10) Press enter , lower left 11) Select units, by using top left scroll buttons. Press enter 12) Position format , select  enter 206  13) Select hddd 0mm‟ss.s” 14) Press Page to get back  B.1.4 After the home visit   Questionnaires: Check questionnaires and home assessment forms for completeness.  Aliquot biological samples: o Breast milk into 6 pre-labelled cryovials and frozen o Nasal samples into 6 pre-labelled cryovials and frozen o Stool sample into 4 pre-labelled cryovials and frozen o Six aliquoted (in the home ) urine cryovials to be frozen Freeze all biological samples at -80 degrees. Fill out HealthDiary forms, including processing and boxing of samples.  Dust samples Enter dust sample collection form into HealthDiary. Leave dust sample at room temperature until shipping every two weeks on a Monday to the Gage Dust Processing Centre in Toronto. Toronto and Vancouver will ship every other Monday 1st and 3rd week of the month. Edmonton and Winnipeg will ship every othe Monday 2nd and 4th week of the month.  Ensure all paperwork for dust sample is complete. DO NOT PROCESS THESE SAMPLES by “accepting these under sample processing”.  See full HealthDiary dust sample Standard Operating procedures for collection and shipping of dust. B.1.4.1 Health Concerns and Precautions  For those coordinators and research assistants that may have allergy/asthma concerns with collecting dust, you may require the use of a mask as follows:   207  A. Educational message The process of vacuuming a home is very important for the study, BUT it can produce a lot of dust. This dust is in the air and easily breathed for approximately one hour after vacuuming. For this reason, we ask that any individuals in the home at the time of sampling who may have asthma be not present in the same room while such sampling is done. If the research assistant has asthma, they, too, can be irritated by the dust from cleaning. A proper dust mask worn over the mouth and nose can reduce this irritation and make cleaning safer. The mask does not eliminate all dusts, but reduces the dust you breathe by at least 95%. You could still have an asthmatic attack from this amount of dust, but it will be less likely than if you do not use the mask. B. How to use a mask Depending upon how dusty the job is, one mask should be usable for 3-5 cleanings. A mask that appears to be dirty should not be used. The mask should fit snugly over the face (beards generally prevent a snug fit). Exhalation should be felt through the coloured button in the front- center of the mask and not around the edges of the mask. The mask should be used during sampling and for about one-half hour afterwards if the technician stays in the rooms that were vacuumed. After use, it should be air-dried in the sun and sealed in a plastic bag between uses. C. Type of mask used: 3M - #8210 - N95  B.1.5 Quality control and quality assurance B.1.5.1 Home visit data collection quality assurance/quality control A. Objective Home assessments performed at age three months (±15 days) for all children born to CHILD Study participants will be monitored for quality control for uniformity in the collection, transport, and storage of environmental data and dust collection. 208  B. Responsabilities of CHILD team Research Coordinator (RC): The Research Coordinator will be responsible for overall coordination of visits, QA/QC for data collection in the field, delivery of dust samples to the CHILD envirobank, and checking data collection forms are complete. The Research Coordinator will provide a monthly report for the national environmental teleconference, and address issues raised by the Research Assistants (RAs) with guidance from their national working group lead as needed. They may request additional resources, such as seeking advice from local health authorities or CMHC consultants, and serve as the liaison with these people. This standard operating procedure (SOP) applies to those members of the CHILD research team and personnel involved in the home visit. An experienced RA is one that has attended 1 or more training sessions and conducted 25 visits. An experienced RA can train other staff members to conduct a home assessment. A trained RA has attended 1 or more training session and conducted 5 supervised visits, and deemed competent by the experienced site trainer. See Figure 1 for certificate of training criteria The Research Coordinator will perform the QA/QC in data collection by shadowing the RAs and collecting a separate questionnaire and RA assessment form in 1 home every 50 home assessments. Data will be double-entered from the paper questionnaire into HealthDiary at the NCC.  Research Assistant (RA) The RA will prepare for and perform the home assessment under the supervision of the research coordinator. They will be responsible for knowing the protocols as outlined in the manual and assuring that all the data are collected, dust samples are properly stored, equipment is collected, and forms are complete at the end of each visit. 209  C. Other CHILD study quality assurance support National environmental assessment Leads The leads will be available as a resource to the team and respond promptly to questions raised by the RC. They will remain in close communication with the site’s Principal Investigators (PIs) and national CHILD Executive and National Coordinating Centre. Communication with experts in the environmental field outside of CHILD will be through the national EWG CHILD leads and the Project Managers of the CHILD study. The EWG leads will maintain communication with the research coordinators through monthly coordinating teleconferences set up by the National CHILD Project Managers, and will be responsible for coordinating the uniform collection of environmental data. These duties include recommending any changes to protocols or data collection instruments for review and final approval by the CHILD Executive. Canada Mortgage and Housing Corporation (CMHC) Canada mortgage and housing consultants may accompany the RAs and RCs on home visits on occasion as part of the QA/QC process, particularly at the beginning of CHILD. They will participate in the monthly teleconferences. Other resources In addition, access is available to local health authorities if issues arise in the assessments that require such attention. CHILD will be enhanced by authorities knowing that we are working in their jurisdiction. Additional expertise will be sought if required from external advisors that can be accessed by the national environmental assessment lead, in collaboration with the CHILD Executive. Each city will have a named Research Coordinator from each city maintain contact with the respective National Environmental Lead. 210  B.1.5.2 Protocol for preparation of ‘QC Samples of Dust’ To be performed once every 12 homes. On average this should work out to be a dust control included in every batch of dust shipped to Gage every two weeks. a. The Research Coordinator/Research Assistant will pre-clean and depyrogenate a stainless steel spatula for transferring NIST dust to a standard dust collection bottle. b. Take the dust collection bottle with a thimble from the supply provided by Gage. This bottle will have a batch ID, but no other labels. c. Add “create a task” in HD to create the control dust sample. d. Print a sample ID control dust label using HealthDiary (“CNDust - 7-xxxxxx0-1”) and affix to the bottle. e. Obtain bottle of NIST (standardized control sample) dust from its storage location (room temperature) and bring to a designated lab bench along with the pre-labelled dust collection bottle and pre-cleaned spatula. f. Open dust collection bottle and place within reach. g. Open NIST dust bottle and using the pre-cleaned spatula, scoop ~200 mg (size of a quarter) of dust and transfer to the dust collection bottle, dumping the NIST dust into the thimble. h. Close NIST dust bottle and the dust collection bottle. i. Place spatula in area where used lab supplies are kept prior to cleaning. j. Clean working space. k. Prepare all other supplies needed for home visit. l. Pack the dust collection bottle with thimble (tightly sealed), now containing NIST dust labelled with “CNDust – 7-xxxxxx-0-1,” along with all the other home visit supplies. m. Bring this bottle to the home(s) visited for that day. 211  n. Open the NIST dust lid for 30 seconds while preparing the vacuum head and placing the thimbles in the hognose. Close lid o. When returning to the lab at the end of the day‟s home visits, this CNDust bottle can be placed in a dust box and information recorded in HD. The time of collection is the same as the first collected dust and the area vacuumed will be entered as “0”. 212  B.2 Figures and supplemental material B.2.1 Figure 1: completed dust collection form (Child’s Bedroom)   213  B.2.2 Figure 2: HealthDiary specimen collection label  ***These labels are 1” by 1” in size   214   B.2.3 Figure 3: worksheet example     

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:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0166377/manifest

Comment

Related Items