TSpace Research Repository tspace.library.utoronto.ca

Kingston Allergy Birth Cohort (KABC);

Exposome Characteristics and Parentally- Reported Respiratory Outcomes to Age 2

Michelle L. North, Jeffrey R. Brook, Elizabeth Y. Lee, Vanessa

Omana, Nadia M. Daniel, Lisa M. Steacy, Greg J. Evans, Miriam Diamond, and Anne K. Ellis

Version Post-print/Accepted Manuscript

Citation (published version)

North M.L., J.R. Brook, E. Y. Lee, V. Omana, N. M. D., L. M., G. J. Evans, M. Diamond and A. K. Ellis “Kingston Allergy Birth Cohort

(KABC); Exposome Characteristics and Parentally-Reported Respiratory Outcomes to Age 2” Ann Allergy, Asthma Immunol 118

(4) 465-473 2017 (IF 2.6) a.

Copyright/License

This work is licensed under the Creative Commons

Attribution 4.0 International License. To view a copy of this license,

visit https://creativecommons.org/licenses/by/4.0/.

How to cite TSpace items

Always cite the published version, so the author(s) will receive recognition through services that track citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace

because you cannot access the published version, then cite the TSpace version in addition to the published version using the permanent URI (handle) found on the record page.

This article was made openly accessible by U of T Faculty.

Please tell us how this access benefits you. Your story matters.

1

1

Kingston Allergy Birth Cohort (KABC); Exposome Characteristics and Parentally-1

Reported Respiratory Outcomes to Age 2 2

3

Michelle L. North 1,2,3,4, Jeffrey R. Brook 5,4, Elizabeth Y. Lee 1, Vanessa Omana 1, Nadia M. 4

Daniel 1,2, Lisa M. Steacy2,3, Greg J. Evans 3,4, Miriam Diamond 7 and Anne K. Ellis 1,2,3* 5

6

1 Department of Biomedical & Molecular Sciences, Queen’s University, Kingston, Ontatio, 7

Canada 8

2 Allergy Research Unit, Kingston General Hospital, Kingston, Ontario, Canada 9

3 Department of Chemical Engineering and Applied Chemistry, University of Toronto, Toronto, 10

Ontario, Canada 11

4 Southern Ontario Centre for Atmospheric Aerosol Research, University of Toronto, Toronto, 12

Ontario, Canada 13

5 Air Quality Research Division, Environment and Climate Change Canada, Toronto, Ontario, 14

Canada 15

6 Division of Allergy and Immunology, Department of Medicine, Queen’s University, Kingston, 16

Ontario, Canada 17

7 Department of Earth Sciences, University of Toronto, Toronto, Ontario, Canada 18

19

Corresponding Author: 20

*Anne K. Ellis, MD, MSc, FRCPC 21

Associate Professor, Division of Allergy & Immunology, Department of Medicine 22

Queen’s University, Doran 1, Kingston General Hospital 23

2

2

76 Stuart Street, Kingston, ON K7L 2V7 24

Telephone: (613) 548-2336, Fax: (613) 546-3079, ellisa@queensu.ca 25

26

Key Words: 27

Allergy, asthma, exposome, smoking, prenatal, epigenetics, mold, environmental health, 28

gestational age, breastfeeding, socioeconomic status, rural health, scented products 29

30

Abbreviations/Acronyms: 31

CI; 95% confidence interval, DOHaD; developmental origins of health and disease, FEV1; 32

forced expiratory volume in 1 second, HR; hazard ratio, KABC; Kingston allergy birth cohort, 33

KGH; Kingston general hospital, LICO; Canadian Low-Income Cut-Off, MNC; mononuclear 34

cells, NCD; non-communicable disease, NHS; national household survey, SES; socioeconomic 35

status. 36

37

Declaration of Sources of Funding 38

MLN; J.A. Stewart Award, Department of Medicine, Queen’s University, Ontario Ministry of 39

Research & Innovation Award, and AllerGen N.C.E. Canadian Allergy and Immune Diseases 40

Advanced Training Initiative (CAIDATI). The study was supported by an Ontario Thoracic 41

Society & Canadian Lung Association Grant-in aid of Research, an Allergy/Asthma Information 42

Association (AAIA) and Canadian Allergy, Asthma & Immunology Foundation (CAAIF) Award 43

for Ontario Research in Food Allergy, and an American College of Allergy, Asthma & 44

Immunology (ACAAI) Young Faculty Support Award. 45

46

3

3

Disclosures/Conflicts of Interest: The authors declare no conflicts related to this 47

manuscript. 48

49

ABSTRACT WORD COUNT: 249 50

MANUSCRIPT WORD COUNT: max 4000 51

TABLES: 3 52

FIGURES: 3 53

54

55

4

4

Abstract 56

Background: The Kingston Allergy Birth Cohort (KABC) is a prenatally-recruited cohort 57

initiated to study the developmental origins of allergic disease. Kingston General Hospital was 58

chosen for recruitment, as it serves a population with notable diversity in exposures relevant to 59

the emerging concept of the exposome. 60

Objective: Establish a profile of the KABC cohort using the exposome framework, and examine 61

parentally-reported respiratory symptoms to age 2. 62

Methods: Data on phase 1 of the cohort (n=560 deliveries) were compiled and multivariate Cox 63

proportional hazard models were used to determine associations with respiratory symptoms. 64

Results: The KABC exhibits diversity within the three exposome domains; “general external” 65

(SES, rural/urban residency), “specific external” (cigarette smoke, breastfeeding, 66

mold/dampness), and “internal” (respiratory health, gestational age). We demonstrate 67

relationships between exposome domains, as residential locale and SES significantly affected 68

characteristics of the home environment. Significant associations emerged between parental 69

reports of wheeze/cough without a cold and exposures within all three domains, including SES, 70

breastfeeding, gestational age, prenatal cigarette smoke exposure, and mold/dampness. 71

Conclusions: The KABC is a unique cohort with diversity that can be leveraged for exposomics-72

based studies, and a notably high prevalence of prenatal smoke exposure. This study 73

demonstrated the impacts of all three domains of the exposome on the respiratory health of 74

KABC children. Ongoing studies using phase 1 of the KABC are further exploring the internal 75

exposome through allergy skin testing and epigenetic analyses, and the specific external domain 76

through in-home environmental analyses, air pollution modeling and ultimately the potential 77

convergence of the two. 78

5

5

Introduction 79

Developmental Origins of Allergic Disease and the Exposome 80

Allergic diseases are common in childhood, and although many risk factors are known, the 81

increase in prevalence over the past quarter century remains poorly understood, and the 82

probability of a substantial environmental contribution has been recognised 1-5. Progress in 83

research on the developmental origins of health and disease (DOHaD), also known as the Barker 84

Hypothesis, has revealed in utero and early-life effects on susceptibility to a wide range of adult 85

diseases, such that it is now generally accepted that many non-communicable diseases, including 86

allergic disease, are likely to arise from cumulative gene-environment interactions 6, 7. The 87

exposome is a new conceptual framework that has been defined to encompass the totality of 88

exposures encountered by an individual throughout the life-course, including the prenatal period 89

8, 9. The exposome framework requires consideration of both the nature of exposures, classified 90

broadly into exposome domains; internal, specific external and general external, as well as their 91

changes and cumulative impact over time 8. The internal domain encompasses host factors, such 92

as sex, age, metabolism, microbiome, epigenetics and inflammatory processes/diseases 8. 93

Specific external exposures are individual-level exposures, such as cigarette smoke and air 94

pollution exposure, while general external exposures encompass wider social, economic and 95

geographic factors, such as socioeconomic status (SES) and urban/rural residence, which also 96

have direct effects on the individual 8. Examining the exposome over time necessitates that birth 97

cohort studies play a key role in exploring this new concept and applying it to gain a deeper 98

understanding of the causes of chronic diseases 9, 10. 99

100

Studying Children with Varied Exposomes to Identify Modifiable Risk Factors 101

6

6

The Kingston Allergy Birth Cohort (KABC) is the first prospective study of allergic disease to 102

be undertaken utilizing the unique catchment area served by Kingston General Hospital (KGH). 103

By offering access to this unique population, with exposome variants that are not as prevalent 104

elsewhere, it compliments larger Canadian cohort efforts such as the Canadian Healthy Infant 105

Longitudinal Development (CHILD) Study 1, 5, 11. As an alternative to recruitment from 106

specialized tertiary or quaternary hospitals, KGH offers access to a community that includes the 107

City of Kingston (a small Canadian city with an approximate population of 152,000), and an 108

approximately equal population residing in surrounding rural areas 12. In addition to coverage of 109

the urban/rural divide in the province of Ontario, the KGH catchment area includes notable 110

diversity in SES. Some Kingston neighborhoods are comprised of households that earn well 111

above the Ontario provincial average, while others are archetypes of multi-generational poverty 112

13, 14. Environmental and prenatal health issues, intimately related to SES, appear to 113

disproportionately affect Kingston, such as prenatal smoking, which at 21.2%, is significantly 114

higher than the rate in Ontario (12.4%), and the nearby city of Ottawa (7.7%) 15. Thus, we chose 115

KGH as the recruitment site in an effort to establish a unique cohort that encompasses exposome 116

variants such as prenatal smoke exposure, while also including urban/rural participants and 117

lower-SES families, in which the prenatal and early life risk factors of allergic disease can be 118

explored. 119

120

The Kingston Allergy Birth Cohort Study Design 121

122

Recruitment for phase 1 of the KABC is complete and follow-up activities are either currently 123

completed or underway as depicted in Figure 1. Briefly, survey data on environment and health 124

7

7

was collected during the third trimester, at 6 months, 1 year and 2 years, and surveys continue to 125

be administered yearly as the children age. Allergy skin testing is being performed on mothers 126

and children. Umbilical cord blood was collected from all available deliveries, and biological 127

samples in the form of blood and plasma are being collected from the children at home visits, 128

concurrent with indoor environmental sampling, and collection of blood and urine from the 129

mothers. An outdoor air pollution sampling campaign was conducted in November/December 130

2014 for the measurement of NO2 and NO for the development of the first land-use regression 16 131

air pollution model of Kingston. Analyses of epigenetic biomarkers in relation to aspects of the 132

prenatal and early life exposome are ongoing. 133

134

Exposomic Analyses of Environmental Exposures and Health Outcomes 135

The purpose of this paper is to characterize prenatal and early-life environmental factors in the 136

KABC through the lens of the exposome, and assess potential associations with parental reports 137

of respiratory symptoms to age 2. Herein we report that both urban/rural residence and SES were 138

associated with prenatal and postnatal environmental exposures. Respiratory symptoms (parental 139

reports of wheeze or cough without a cold) were significantly associated with prenatal cigarette 140

smoke exposure, lower SES and the presence of mold in the home during early life. Increasing 141

gestational age at birth, and breastfeeding for 6 months or more were negatively associated with 142

early-life respiratory symptoms. 143

144

Methods 145

146

Recruitment and Umbilical Cord Blood Collection 147

8

8

Participants had to be able to communicate in English, and provide written informed consent 148

prior to any study-specific procedures. Ethical clearance for the study was provided by the 149

Queen's University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board. 150

Potential participants of the KABC were pregnant women, aged 18 years or older, in the second 151

or third trimester. Participants were eligible if they intended to deliver at KGH, did not have pre-152

eclampsia or insulin-dependent diabetes, intended to retain custody of the child, were not 153

intending to bank their umbilical cord blood, and did not exhibit any other risk factors that their 154

physician felt would preclude their safe participation. Multiple births and consecutive siblings 155

were not excluded. Participants were recruited via posters at family physician’s offices, 156

midwives clinics, obstetrics clinics and at KGH. Potential participants were also invited to 157

participate at clinic appointments, prior to scheduled Cesarian sections, or in early labor, if they 158

responded positively to a query from a healthcare provider regarding their willingness to speak 159

with study staff. Umbilical cord blood was collected in heparinized syringes at delivery. Cell 160

types of interest were isolated fresh whenever possible and excess cord blood was processed to 161

mononuclear cells for archiving in liquid N2 in freezing medium. Umbilical cord blood was not 162

collected from approximately 26% of women who consented to the study (Figure 1) due to 163

failure to identify the woman as a study participant, cord blood loss from tearing of the cord or 164

placenta or other factors surrounding the delivery. 165

166

Prenatal and Postnatal Surveys 167

A prenatal survey was administered at enrolment. The first 51 participants to enroll in the study 168

were administered a less-detailed pilot survey. The pilot survey included questions regarding 169

parental allergic disease, and maternal exposure to cigarette smoke. The full survey was 170

9

9

administered to all women who enrolled subsequently, and included additional questions 171

regarding the home and other environmental exposures. Efforts were made to harmonize the 172

information gathered in the full survey with the CHILD study, by adapting questions from 173

Takaro and colleagues 1, while limiting survey length/burden on the participants as much as 174

possible. 175

176

Follow-up postnatal data collection is continuing, and the analyses presented herein focus on 177

surveys administered at 6 months, 1 year, and 2 years following delivery, by mail or in person. If 178

a child moved during the first 2 years of life it was determined whether or not the move changed 179

the child’s classification in terms of any variables, and if so, the child was included under the 180

classification where they spent 60% or more of their time under the age of 2. If that criteria was 181

not able to be met, the status of the child with respect to that exposure was considered missing. 182

Definitions of prenatal and postnatal variables are given in Table 1. 183

184

SES Estimation Using National Household Survey (NHS) Data 185

Addresses were used to assign National Household Survey (NHS) 17 data to each participant, by 186

census-tract. If tract-level data was not available or censored for the neighborhood, the closest 187

adjacent tract was used if the distance was less than 2 kilometers. If this was not possible, such as 188

in rural areas, data for the rural town of residence, or census division were used. 189

190

Definition of Early Life Respiratory Symptoms 191

10

10

When reporting respiratory symptoms, parents were asked to only include wheezing and 192

coughing not associated with a cold. If they reported either respiratory symptom they were asked 193

how old the child was (in months) when the coughing/wheezing started. 194

195

Statistical Methods 196

Percentages of the cohort were calculated based on all non-missing for that category. We aimed 197

to examine the correlation structure of the exposome of KABC children during the in utero 198

period and early life, using the variables in Table 1. Thus, we computed “exposome globes” 199

separately for each of those developmental time periods. The exposome globe is a technique 200

similar to that of the “relevance network” to find correlations between expressed genes 18. Patel 201

and Manrai recently developed the relevance network technique into the exposome globe to aid 202

in visualizing and understanding relationships between various exposures in exposomics 18. We 203

computed the non-parametric correlation coefficients and p values between each pair of 204

environmental factors in R and generated exposome globes using Circos to visualize the top 205

quartile of correlation coefficients 19, 20. P values were adjusted post-hoc using the Benjamini-206

Hochberg false discovery rate for multiple tests of significance in R 19, 21. 207

208

Correlations between exposures were used to inform our subsequent examination of factors 209

associated with parental reports of wheeze/cough without a cold. We employed a stepwise 210

analysis; we first explored the potential association between each exposure and respiratory 211

outcomes using univariate Cox proportional hazard models. All variables that exhibited p < 0.15 212

upon univariate analysis plus any variables they were significantly associated with in the 213

previous exposome globe analyses (Supplemental Tables 1 & 2), were entered into a 214

11

11

multivariate Cox proportional hazard regression model. The large number of significant 215

relationships between the variables resulted in the majority of variables being included, thus the 216

inclusion of variables was further optimized using Akaike information criterion (AIC). Prism 217

version 6.0 (Graphpad Software Inc.) was used for graphing Kaplan–Meier curves and associated 218

Fisher’s exact/Chi squared tests, while R was used for all other analyses R 19. Two-tailed p 219

values < 0.05 were considered significant. 220

221

Results 222

223

Characteristics of the KABC Cohort 224

First, we set out to understand the basic exposome characteristics of the KABC participants at 225

enrolment by collating survey data (Table 1). The general external exposures of residential 226

setting and SES demonstrated diversity, as residential setting was almost evenly split between 227

urban and rural, and incomes ranged from below the Canadian Low-Income Cut-Off (LICO), to 228

well above the provincial average 22. More than a quarter of children were exposed to cigarette 229

smoke in utero, and prenatal smoke exposure was an everyday occurrence for more than 60% of 230

the children whose mothers reported being exposed to smoke (Table 1). The diversity in 231

residential exposures reported in the KABC was also revealed the the broad range of potential 232

pollution sources; the most common potential source within 100 meters of the home was a major 233

road, while the second most common was a farm. Potential exposures within the home also 234

varied, as more than two thirds of the cohort reported furry pets living in the home, one fifth 235

reported mold or dampness in the home, and approximately half reported using air fresheners at 236

least once a week (Table 1). 237

12

12

238

Follow-up Survey Responders Recapitulate the Diversity of the Cohort as a Whole 239

During the first 2 years of the study, 3% of women withdrew and 16% moved or otherwise 240

became unreachable by phone, post and email (Figure 1). Of the 452 mothers remaining, 52% 241

responded to at least one survey during the first 2 years of follow-up. The remaining 48% were 242

non-responders for the purposes of this analysis (age 2 and under) but were not lost to follow-up, 243

as they may contribute data to future analyses. Thus the prenatal survey analyses includes data 244

from 560 children, while the postnatal includes 235. There were no significant differences 245

between the whole cohort and those that responded to surveys in terms of urban/rural residence, 246

SES, or proportions of mothers who self-reported having allergic disease (p>0.05, Fisher’s Exact 247

test). 248

249

Correlation Structure of Prenatal and Postnatal Exposures 250

Characteristics of the home environment were assessed as part of the prenatal survey, and we 251

observed strong evidence of interconnectedness between several individual factors (Figure 2A). 252

SES, housing type, garage type and urban/rural residence were the factors investigated that were 253

most highly linked to other aspects of the prenatal environment, with 8, 12, 9 and 7 significant 254

associations with other exposures, respectively (adjusted p<0.05, Supplemental Table 1). 255

Prenatal smoke exposure was significantly associated with housing type, home age, and the 256

regular use of candles and air fresheners in the home (adjusted p<0.05, Supplemental Table 1). 257

The presence of a dog in the home was associated with housing type and rural residence, while 258

the presence of a cat was not significantly associated with any other home environmental factors 259

(adjusted p<0.05, Supplemental Table 1). In addition to more indoor dogs, rural residence was 260

13

13

also significantly associated with a greater proportion of single family homes, wood heating, 261

living near a farm, and higher SES, while it was negatively associated with living near a major 262

road or parking lot (adjusted p<0.05, Supplemental Table 1). While the top 25% of correlation 263

coefficients demonstrated connectedness of maternal and paternal allergy to other factors, they 264

were not significantly associated with any other prenatal characteristics (Supplemental Table 265

1). 266

267

The interconnectedness of the “web” of environmental factors surrounding each child were also 268

evident among the postnatal characteristics examined (Figure 2B). Similar to the prenatal time 269

period, the postnatal factors that correlated significantly with the largest numbers of other factors 270

included housing type, urban/rural status, and SES (Supplemental Table 2). Additional 271

relationships emerged, such as significant associations between Cesarean section delivery, the 272

presence of older siblings in the home and gestational age (adjusted p<0.05, Supplemental 273

Table 2). Breastfeeding was associated with significantly lower rates of air freshener and 274

candle/incense use (adjusted p<0.05, Supplemental Table 2). Home age was also significantly 275

related to SES and newer homes contained significantly less carpeting (adjusted p<0.05, 276

Supplemental Table 2). The exposome globe analysis and correlation analyses also revealed 277

that KABC children growing up in lower SES homes were also significantly more likely to be 278

exposed to mold in the home (adjusted p<0.05, Supplemental Table 2). 279

280

Prenatal Factors Associated with Rates of Parentally-Reported Respiratory Symptoms 281

Examination of correlation structures in exposome studies is meant to promote transparency and 282

facilitate the interpretation of exposure effects in light of their relationships to other exposures 283

14

14

(potential confounders) 8, 23. Thus, we considered the interrelatedness of exposures as we went on 284

to examine factors associated with parental reports of wheeze/cough without a cold. In our step-285

wise analysis, the prenatal presence of a cat in the home, the regular use of air fresheners, SES, 286

prenatal smoke exposure, home age and garage type met the a priori criteria for inclusion in the 287

multivariate model. These factors plus all variables that demonstrated significant associations 288

with them in the exposome globe analysis were included in an AIC analysis for selection of the 289

optimal model while minimizing loss of information. In the final multivariate model, adjusting 290

for SES, cat in the home, gas heat, the use of air fresheners, and carpeting, only prenatal smoke 291

exposure was significantly associated with parental reports of cough or wheeze without a cold to 292

age 2 (adjusted p = 0.00006, Table 2). The overall incidence rate of parental reports of 293

respiratory symptoms was 0.19 cases/person-year. 294

295

Postnatal Factors Associated with Rates of Parentally-Reported Respiratory Symptoms 296

297

In univariate Cox proportional hazard models of postnatal variables, gestational age, SES, indoor 298

cat exposure, breastfeeding for 6 months or more, mold in the home, postnatal smoke exposure, 299

and the regular use of air fresheners and candles met the criteria for potential inclusion in the 300

multivariate model. All postnatal characteristics that significantly correlated with these variables 301

of interest in the exposome globe analysis were also included (Supplemental Table 2). After 302

AIC optimization of the model, increased gestational age at birth, the presence of siblings in the 303

home and breastfeeding for at least 6 months were found to be significantly negatively associated 304

with respiratory symptoms (adjusted p < 0.05, Table 3). The frequent use of air fresheners in the 305

home (solid, plug-in or spray at least once a week) and the presence of mold in the home were 306

15

15

significantly associated with increased parental reports of wheeze or cough without a cold during 307

the first 2 years of life (P<0.05, Table 3). Model effect estimates and p values for each variable 308

of interest were adjusted for all other variables included in the multivariate model, including 309

gestational age, breastfeeding, siblings, air fresheners and mold. 310

311

As the internal exposome characteristic of increased gestational age at birth was significantly 312

associated with lower risk of respiratory symptoms, we further sought to determine if this effect 313

was driven by children born prematurely. We excluded the 13 children who were born at <37 314

weeks gestation and found that the association remained significant in both univariate (hazard 315

ratio = 0.74, p = 0.022) and multivariate models (hazard ratio = 0.75, adjusted p = 0.029). Thus, 316

increasing gestational age is significantly associated with lower rates of parental reports of 317

respiratory symptoms before the age of 2 in the KABC, across all births, as well as in term 318

babies only. 319

320

Prenatal and Postnatal Factors Identified Demonstrate Effects on Symptom Development on 321

Kaplan–Meier Curve Analyses 322

323

We visualized the effects of the variables found to have significant effects on respiratory 324

symptom development using Kaplan–Meier curves (Figure 3). The curve of symptom 325

development for prenatal smoke-exposed children was significantly different from unexposed, 326

by the Log-rank Mantel-Cox test (p = 0.008), and the exposed/unexposed curves significantly 327

departed from one another at 3 months of age (p = 0.008, Fisher’s exact test). The Kaplan–Meier 328

curves of air freshener and mold exposure were also found to be significantly different from the 329

16

16

curves of their respective unexposed groups (p = 0.0065 and p = 0.01, respectively, by the Log-330

rank Mantel-Cox test). The symptom-development curve of children exposed to the regular use 331

of air fresheners departed significantly from control at 12 months (p = 0.009, Fisher’s exact test), 332

while mold-exposed children departed from control at 3 months (p = 0.006, Fisher’s exact test). 333

Factors found to be protective against symptom development also demonstrated significant 334

effects in Kaplan–Meier analyses (p = 0.02 for gestational age and p = 0.001 for breastfeeding), 335

and departure from controls at 3 months (p = 0.04), and 9 months (p = 0.02). Kaplan–Meier 336

analysis of the effect of siblings did not demonstrate significance. 337

338

17

17

Discussion 339

340

Using the Exposome Framework in Non-Communicable Disease Research 341

The purpose of the exposome concept is to improve identification of environmental risk factors 342

in epidemiological studies and potentially gain new insight into critical windows of vulnerability 343

throughout the life course 8. The examination of the pregnancy exposome and the development 344

of the exposome globe concept have begun to provide early insights 18, 23, and new techniques for 345

data analysis in exposomics are ardently being developed and tested 24. Large consortia studies, 346

such as the human early-life exposome (HELIX) collaboration of 6 European cohorts are using 347

“big health data” and new techniques such as smartphone applications to take important steps 348

towards measuring the life-course exposome 25. However, there are also significant challenges in 349

applying the exposome concept and subsequently acting upon these findings to reduce individual 350

or population risk. Here we have attempted to align environmental exposures within a new birth 351

cohort with this concept to begin to turn it into practice by examining childhood cough and 352

wheeze for the first time through the lens of the exposome. We hypothesize that starting with 353

birth cohorts, such as the KABC, provides the greatest opportunity to fully embrace and test the 354

concept, yet considerable challenges can be expected as the exposome approach evolves. 355

356

Non-communicable diseases (NCDs) have been declared to be a worldwide public health crisis 357

26, and allergic diseases, including asthma, are the most common NCDs of childhood 27. In this 358

study we analyze longitudinal survey data from the third trimester to 2 years of age using the 359

exposome globe technique, and we examine parentally-reported respiratory health outcomes. In 360

ongoing follow-up studies we are making more in-depth measures of the specific external 361

18

18

exposome, including environmental sampling in the bedrooms of the children, and measures of 362

the internal exposome, such as genome-wide epigenetic biomarkers in umbilical cord blood and 363

peripheral blood. Using the exposome framework may lead to a better understanding of 364

modifiable risk factors as well as the basic mechanisms underlying the effects of these 365

exposures, individually and collectively. 366

367

Links Between the Three Domains of the Exposome 368

Using the KABC cohort, we examined relationships between the three domains of the exposome; 369

internal, specific external, and general external. We found that the general external domain 370

factors of SES and urban/rural status were strong modifiers of the specific external exposome, as 371

they were significantly associated with multiple exposures such as housing type, age of the 372

home, indoor dog ownership, wood heating, proximity to traffic pollution and even consumer 373

product behaviors such as the use of air fresheners in the home. 374

375

The use of the exposome globe concept brought to light some interesting relationships that are 376

sensible, but may not have been recognized otherwise, such as the associations between 377

Cesarean section, the presence of older siblings in the home and gestational age. This is likely 378

due to repeat elective Cesarean sections following the previous Cesarean delivery of an older 379

child, and the scheduling of the surgery in advance of the due date 28. This is a fascinating 380

example of how an individuals’ exposome extends to the pre-conception stage, as the mother’s 381

previous pregnancy history is linked to the subsequent child’s mode of delivery and gestational 382

age; factors that have the potential to change their disease risk across their life course. 383

384

19

19

In the analysis of parentally-reported respiratory symptoms, the importance of links between 385

different exposures and exposome domains remained evident. Inclusion of other factors known 386

to be associated with the predictors of interest improved our ability to ensure we were controlling 387

for as many potential confounders as possible. Thus, the exposome concept helped to organize 388

our thinking around risk factors in the KABC cohort; how they are related and influence each 389

other. As follow-up continues and biological samples are analyzed, the exposome concept may 390

aid in identifying novel risk factors, biomarkers of exposures, or exposure-classes. 391

392

Gestational Age and Parental Reports of Wheeze/Cough in Early Life 393

As a host factor that we carry with us for our entire lives, gestational age at birth is an important 394

aspect of the internal exposome. It is known that preterm birth is the leading cause of child death 395

in high-income and middle-income countries, and the USA is one of the ten countries with the 396

highest numbers of preterm births 29. The percentage of children in the KABC who were born 397

preterm was consistent with the overall 4.9% rate of preterm birth in Canada 30. However, the 398

negative association between weeks of gestation and reports of respiratory symptoms in the 399

KABC was not limited to preterm deliveries. Many follow-up studies of lung function focus on 400

children born very preterm (<32 weeks), and demonstrate persistent negative effects 31. Recently, 401

children in the 402

Children Allergy Milieu Stockholm Epidemiological Survey (BAMSE) cohort born moderate-to-403

late preterm were examined, revealing that these groups also experience significant decrements 404

in forced expiratory volume in 1 second (FEV1) 32. Relatively little research has focused on the 405

effects of gestational age on respiratory health in term infants, but one study recently 406

demonstrated an increased risk of asthma in early-term births (37-38 weeks, adjusted odds ratio, 407

20

20

1.2; 95% CI, 1.1-1.4), and a decreased risk of asthma in those born at 41 weeks or later (adjusted 408

odds ratio, 0.9; 95% CI, 0.8-1.0) 33. This is consistent with our findings, suggesting that increases 409

in gestational age across the entire spectrum may be beneficial in terms of reduced risk of 410

respiratory disease. 411

412

Maternal Smoking During Pregnancy 413

Maternal smoking is a crucial modifiable prenatal risk factor that is known to profoundly affect 414

the health of children. A recent meta-analysis of prospective cohorts determined that prenatal 415

smoke exposure was significantly associated with incident asthma 34. While we are not able to 416

determine incident asthma from the parental-report data that is the focus of this analysis, the 417

significantly higher rate of wheeze/cough without a cold in the KABC children who were 418

exposed prenatally is consistent with the previous meta-analysis of Burke et al. 419

420

While we focused on one area known to be affected by high maternal smoking rates, there are 421

many other pockets of high-risk populations across North America. Geographically, clusters of 422

higher than average maternal smoking rates have been shown to exist in the American Midwest 423

35, and Western Canada 36. Other factors such as economic-vulnerability 35, 37, education 38, 424

maternal age less than 24 35, 36, 38, cultural background 39, and experiencing stressful events 425

before/during pregnancy 40 are all associated with increased risk of maternal smoking during 426

pregnancy. While smoking rates in the United States have fallen over the past 50 years, during 427

the past 10 years tobacco-control efforts have been insufficient to reach national objectives of 428

reducing prenatal smoking 38, 41. Thus, maternal smoking during pregnancy remains a pressing 429

public health issue. A spatiotemporal external exposome framework has previously been used to 430

21

21

examine “hotspots” where the prevalence of low birth weight was higher than expected in the 431

United States 42. While Kingston was a known “hotspot” for maternal smoking that was targeted 432

for this study, applying a longitudinal spatial exposome analysis applied to maternal smoking 433

may provide new knowledge for targeted interventions in the future. 434

435

Breastfeeding and Siblings 436

In this study, both breastfeeding and the presence of siblings in the home were associated with a 437

lower incidence of parentally reported respiratory symptoms. Breastfeeding during early life is 438

an important aspect of the specific external exposome, which may have long-term effects on the 439

internal exposome, such as aiding in the establishment a healthy gut microbiome, as well as 440

epigenetic and metabolomic effects 11, 43-45. Siblings have also been shown to affect gut 441

microbiota development in early life and previously related to rates of asthma and allergies 46, 47. 442

443

The rate of breastfeeding reported herein is comparable to the rate of 53.1% determined for non-444

exclusive breastfeeding to at least 6 months by the Canadian Community Health Survey 48. The 445

relationship between breastfeeding and the risk of childhood asthma has been studied 446

extensively, but remains controversial. A number of studies have demonstrated protection, while 447

others report no effect, or even increased risk among breastfed children, particularly if the 448

mother is atopic 49-51. In the KABC we found a negative association between breastfeeding for at 449

least 6 months and reports of wheeze or cough without a cold. There was no difference between 450

allergic and non-allergic KABC mothers in terms of breastfeeding rates, or reports of respiratory 451

symptoms in their children. Thus in this cohort, breastfeeding appears to have positive effects 452

towards respiratory health outcomes in the children of both atopic and non-atopic mothers. 453

22

22

Although some studies fail to find a specific effect of breastfeeding on rates of asthma, it is 454

widely recognized that breastfeeding provides a range of benefits for growth, immunity, and 455

development 44, 52. The prospective data collected by the KABC on breastfeeding duration and 456

solid food introduction will be useful for further studies on the developmental origins of 457

respiratory health when skin testing, epigenetic and pulmonary function data become available 458

on these children. 459

460

Indoor mold and scented product exposures 461

Airborne specific external exposures in the home, such as indoor mold and air fresheners may be 462

particularly relevant to respiratory health in early life. In cross-sectional studies and meta-463

analyses, mold exposure has been linked to increased risk of allergic respiratory health outcomes 464

53, 54. Beta-glucan, a polymer present in fungal cell walls, was demonstrated to be associated with 465

persistent atopic asthma and bronchial hyper-responsiveness in another Canadian cohort 55. Our 466

results are in line with these findings, as we demonstrate that the strongest postnatal association 467

between increased reports of wheeze/cough and any variable was with mold/dampness in the 468

home. 469

470

The use of scented products in the home is another emerging indoor air quality issue. A recent 471

study demonstrated that in addition to perfume-related chemicals, scented products also contain 472

high levels (>1000μg/g) of phthalates 56. Links have been made between exposure to phthalates 473

and childhood asthma, however, further longitudinal studies are needed to determine causality 57. 474

Phthalate levels will be measured in samples taken during home visits in the KABC cohort and 475

23

23

correlations between health outcomes and epigenetic effects in blood samples collected 476

concurrently may emerge. 477

478

The Time-Integrated Exposome 479

By definition, the exposome changes over time based on cumulative exposure. We found that 480

different groups of variables behaved differently in terms of their level of continuity between the 481

prenatal and postnatal periods examined. For example, urban/rural residence, housing type, and 482

SES did not change unless a child moved into a new home that changed their classification in 483

terms of those variables. Less than 9% of the children displayed discordance between those 484

prenatal and postnatal variables. While other variables, such as mold in the home, the use of air 485

fresheners and smoke exposure have the prospect of changing between the prenatal and postnatal 486

periods without the family moving. For example, air fresheners and mold exposure each 487

demonstrated discordance in approximately 30% of children from the prenatal to postnatal 488

periods. Both of these exposures were found to be significantly associated with respiratory 489

symptoms when the exposure occurred postnatally. This may point to biological mechanisms 490

such as placental protection and/or a mechanism that is dependent on direct exposure of the 491

child’s lungs. 492

493

Smoking provided another example of the importance of the time-integrated exposome. While 494

the total percent of the cohort exposed to cigarette smoke did not change dramatically from the 495

prenatal (26%) to postnatal (23%) periods, the degree of smoke exposure of the children 496

appeared to change significantly, from 16% of prenatal smoke being an everyday occurrence, to 497

only 2% of children who were reported to be exposed to smoke in the home after birth, with the 498

24

24

remaining 21% representing third-hand or second-hand somewhere other than the home. In 499

addition, the parents of 32 children (almost 14% of survey responders) who did not report any 500

smoke exposure during pregnancy, did report postnatal smoke exposure, thus some children 501

changed exposome characteristics dramatically. The more chronic and consistent nature of the 502

prenatal smoke exposure, or the developmental timing, or both, may play roles in our finding that 503

prenatal smoke exposure was significantly related to respiratory symptoms to age 2, while 504

postnatal smoke was not. It is possible that with a larger sample size, or with additional data 505

collection later in life, the effects of postnatal smoke exposure will become evident. Thus the 506

effects of several exposures found to affect respiratory health in this study may be as a result of 507

the time integrated pre-natal exposure or the time-integrated exposure during infancy, or both. 508

509

Time to Onset of Symptoms 510

511

The postnatal time point at which the Kaplan–Meier curves of the exposed children departed 512

from the unexposed also revealed some intriguing differences between the exposome 513

characteristics of interest. The effects of prenatal smoke exposure, gestational age and postnatal 514

mold exposure were significant by 3 months of age. Both prenatal smoke exposure and 515

gestational age have been associated with potentially-permanent changes in lung development 58. 516

While for mold exposure one explanation might be immediate irritating effects upon inhalation 517

of mold in infants, although the indoor microbial environment is complex and additional studies 518

will be needed to further assess mechanisms 54. The Kaplan–Meier curves of children who were 519

not breastfed for at least 6 months and those who were exposed to the regular use of air 520

fresheners indoors significantly diverged from control at 9 months and 12 months, respectively. 521

25

25

The 9 month time point at which the effects of breastfeeding became apparent may be due to the 522

time it takes for the establishment of the microbiome and other immune-modulating factors 523

associated with breastfeeding 11, 43-45. Air fresheners represent a complex mixture of chemicals 524

with various toxicological profiles 59. Thus, the differences in the onset of symptoms of each 525

group of children with the exposome characteristics of interest may reflect the differences in the 526

impacts of their time integrated exposures. 527

528

Limitations of the Kingston Allergy Birth Cohort (KABC) 529

The generalizability of studies is known to depend heavily on both the population and the 530

sampling scheme. Some of our recruitment was carried out through posters at family physician’s 531

offices, midwives clinics, and KGH. It is possible that the participants who contacted us through 532

this route were interested because someone in their family had been affected by allergic disease. 533

However, ~90% of those in the study were approached in person at appointments, prior to 534

scheduled Cesarian sections, or in early labor, and 79% of women who were asked to participate 535

in the study consented. Thus, we acknowledge that there may be some participation bias, but 536

conclude that it is not likely to have caused overwhelming deviations from the general 537

population. However, as study staff regularly attended the labor and delivery ward prior to 538

scheduled Cesarean sections to ask if women may be interested in the study, our study 539

population does encompass a higher than average number of Cesarean section births (53% in the 540

KABC compared to 25.9% in the public health region) 15. The percentage of smoking mothers in 541

the KABC is consistent with recent public health reports for the region, suggesting that the 542

smoking population in Kingston was not underrepresented 15. Representation from both the 543

highest and lowest SES census tracts also demonstrates that the cohort adequately captured the 544

26

26

notable diversity of SES of the Kingston area. While the percentage of eligible women in the 545

cohort who completed follow-up surveys was lower than ideal, at just over 50%, the diversity in 546

the sample in terms of prenatal smoking and SES remained. Some women who did not respond 547

to surveys to age 2 have already provided data for the ongoing collection at age 3, and attempts 548

at follow-up will continue. 549

550

In this study we did not find an association between self-reported proximity to traffic pollution 551

and respiratory symptoms in the children. However, studies have found that self-reported air 552

pollution exposure is only weakly associated with modeled air pollution exposure 60. In the 553

CHILD study, postnatal exposure to traffic-related air pollution was associated with atopy at 1 554

year of age, using models that incorporated temporal variability and residential mobility 61. We 555

are currently working to quantify measures of the specific external exposure of outdoor air 556

pollution at the home address through the development of a traffic-related air pollution model for 557

Kingston based upon a short-term campaign of NO2 and NOx measurement. Particularly because 558

of the inclusion of both urban and rural participants in the KABC, significant gradients in 559

exposure may be present, and further studies may be able exploit this pattern. 560

561

Another limitation to the study is that, save for SES, we did not collect measures of the prenatal 562

and postnatal psychosocial environment and stress, which may be involved in child development 563

and the prevalence of allergic and other diseases 62. The KABC was designed for recruitment to 564

be carried out in phases, and plans are being made to collect more in-depth psychosocial health 565

and stress exposure data in phase 2. Further phases will also increase the overall sample size and 566

27

27

provide the opportunity to validate the associations made herein, while delving deeper into 567

potential biological mechanisms. 568

569

Conclusions 570

The present study introduces a new cohort with a high prevalence of in utero exposure to 571

cigarette smoke, drawn from a mixed population of small-city urban and rural residents, with 572

diverse socioeconomic status. This cohort has the advantage of exhibiting diversity within the 573

three exposome domains. We demonstrated interplay between the general and specific exposome 574

domains, as urban/rural status and SES was demonstrated to significantly affect characteristics of 575

the home environment. This is the first to examine the development of cough and wheeze 576

without a cold in early childhood through the lens of the exposome. We report increased 577

incidence associated with prenatal cigarette smoke exposure, mold/dampness in the home and 578

the use of air fresheners. Breastfeeding for 6 months or more and increasing gestational age were 579

associated with decreased respiratory reports. These findings will be followed by further 580

investigations of the home environment, and the internal exposome including epigenetics and 581

metabolomics. We will continue to approach the cohort from the exposome perspective and 582

attempt to advance that thinking further for KABC and the ‘exposome community’ in general. 583

28

28

Figure Legends 584

585

Figure 1: Flow chart of Kingston Allergy Birth Cohort (KABC) enrolment, withdrawal, loss to 586

follow-up and follow-up survey completion. 587

588

Figure 2: Exposome globes depicting the relationships, in terms of non-parametric correlation 589

coefficients, between components of the A) prenatal and B) postnatal environments. 590

591

Figure 3: Kaplan–Meier curves depicting the effects of A) prenatal smoke, B) the regular use of 592

air fresheners in the home, C) postnatal mold exposure in the home, D) gestational age, E) 593

breastfeeding for at least 6 months, and F) the presence of siblings in the home on parental 594

reports of symptoms of wheeze and cough without a cold. 595

596

597

Acknowledgements 598

KABC Investigators also include Dr. Michael Kobor, Dr. Meaghan Jones. The study staff would 599

like to thank the families who took part in the study and the doctors, midwives and nurses at 600

Kingston General Hospital. Thank you to Mena Soliman, Jenny Thiele, Michelle Roddy, Roberta 601

Faroldi, Kiyoka Sasaki, Sara Angélica Urquieta Ruelas, Katie Corscadden, Ibrahim Sandoqjy, 602

Julianne Murphy, Nancy Rogers, Katrina Au, Kerolyn Shairsingh, Dr. Mohammad Alavinia, 603

Yuchao Wan, Congqiao Yang and Professor Marianne Hatzopoulou for their assistance on 604

various aspects of the KABC. 605

29

29

References (max 60) 606

607

1. Takaro TK, Scott JA, Allen RW, Anand SS, Becker AB, Befus AD, et al. The Canadian 608 Healthy Infant Longitudinal Development (CHILD) birth cohort study: assessment of 609 environmental exposures. J Expo Sci Environ Epidemiol 2015. 610

2. Arrandale VH, Brauer M, Brook JR, Brunekreef B, Gold DR, London SJ, et al. Exposure 611 assessment in cohort studies of childhood asthma. Environ Health Perspect 2011; 612 119:591-7. 613

3. Grammatikos AP. The genetic and environmental basis of atopic diseases. Ann Med 614 2008; 40:482-95. 615

4. Neeland MR, Martino DJ, Allen KJ. The role of gene-environment interactions in the 616 development of food allergy. Expert Rev Gastroenterol Hepatol 2015:1-8. 617

5. Subbarao P, Anand SS, Becker AB, Befus AD, Brauer M, Brook JR, et al. The Canadian 618 Healthy Infant Longitudinal Development (CHILD) Study: examining developmental 619 origins of allergy and asthma. Thorax 2015; 70:998-1000. 620

6. Barker DJ, Thornburg KL. The obstetric origins of health for a lifetime. Clin Obstet 621 Gynecol 2013; 56:511-9. 622

7. North ML, and Ellis, A.K. . The role of epigenetics in the developmental origins of 623 allergic disease. . Ann Allergy Asthma Immunol 2011; 106(5):355-61. 624

8. Wild CP. The exposome: from concept to utility. Int J Epidemiol 2012; 41:24-32. 625 9. Wright ML, Starkweather AR, York TP. Mechanisms of the Maternal Exposome and 626

Implications for Health Outcomes. ANS Adv Nurs Sci 2016; 39:E17-30. 627 10. Vrijheid M. The exposome: a new paradigm to study the impact of environment on 628

health. Thorax 2014; 69:876-8. 629 11. Azad MB, Konya T, Maughan H, Guttman DS, Field CJ, Chari RS, et al. Gut microbiota 630

of healthy Canadian infants: profiles by mode of delivery and infant diet at 4 months. 631 CMAJ 2013; 185:385-94. 632

12. Kingston Co. City of Kingston Statistical Profile. Planning and Development, 633 Kingston, Ontario May 2008. 634

13. A Socio-Economic Profile of Kingston, Ontario. http://www.ncronline.ca; Natural 635 Capital Resources. 636

14. Kingston Co. Ready To Do Better. The Mayor’s Task Force on Poverty 2007; 637 http://vibrantcanada.ca/files/kingston_poverty_task_force_report.pdf. 638

15. Perinatal Health Report 2008. . BORN Ontario 2010; Eastern Ontario Public Health 639 Region. 640

16. Oiamo TH, Luginaah IN, Buzzelli M, Tang K, Xu X, Brook JR, et al. Assessing the 641 spatial distribution of nitrogen dioxide in London, Ontario. J Air Waste Manag Assoc 642 2012; 62:1335-45. 643

17. Canada S. National Household Survey User Guide. National Household Survey, 2011 644 2015; Catalogue no. 99-001-x2011001:http://www.statcan.gc.ca. 645

18. Patel CJ, Manrai AK. Development of exposome correlation globes to map out 646 environment-wide associations. Pac Symp Biocomput 2015:231-42. 647

19. Team RC. R statistical functions. R Foundation for Statistical Computing, Vienna, 648 Austria. 2016; R-core@r-project.org. 649

30

30

20. Krzywinski M, Schein J, Birol I, Connors J, Gascoyne R, Horsman D, et al. Circos: an 650 information aesthetic for comparative genomics. Genome Res 2009; 19:1639-45. 651

21. Benjamini Y, Hochberg, Y. Controlling the false discovery rate: A practical and 652 powerful approach to multiple testing. Journal of the Royal Statistical Society. Series 653 B. 1995; 54:289-300. 654

22. Division IS. Income Research Paper Series. Low Income Lines, 2013-2014: Update 655 2015; Statistics Canada:http://www.statcan.gc.ca/pub/75f0002m/75fm2015002-656 eng.pdf. 657

23. Robinson O, Basagana X, Agier L, de Castro M, Hernandez-Ferrer C, Gonzalez JR, et 658 al. The Pregnancy Exposome: Multiple Environmental Exposures in the INMA-659 Sabadell Birth Cohort. Environ Sci Technol 2015; 49:10632-41. 660

24. Agier L, Portengen L, Chadeau-Hyam M, Basagana X, Giorgis-Allemand L, Siroux V, et 661 al. A Systematic Comparison of Linear Regression-Based Statistical Methods to 662 Assess Exposome-Health Associations. Environ Health Perspect 2016. 663

25. Vrijheid M, Slama R, Robinson O, Chatzi L, Coen M, van den Hazel P, et al. The human 664 early-life exposome (HELIX): project rationale and design. Environ Health Perspect 665 2014; 122:535-44. 666

26. Beaglehole R, Bonita R, Horton R, Adams C, Alleyne G, Asaria P, et al. Priority actions 667 for the non-communicable disease crisis. Lancet 2011; 377:1438-47. 668

27. Bousquet J. Global Surveillance, prevention and control of chronic respiratory 669 disease: a comprehensive approach. Geneva, World Health Organization. 2007; 670 Bousquet, J. and Khaltaev, N. 671

28. Hansen AK, Wisborg K, Uldbjerg N, Henriksen TB. Elective caesarean section and 672 respiratory morbidity in the term and near-term neonate. Acta Obstet Gynecol 673 Scand 2007; 86:389-94. 674

29. Blencowe H, Cousens S, Oestergaard MZ, Chou D, Moller AB, Narwal R, et al. 675 National, regional, and worldwide estimates of preterm birth rates in the year 2010 676 with time trends since 1990 for selected countries: a systematic analysis and 677 implications. Lancet 2012; 379:2162-72. 678

30. Garn JV, Nagulesapillai T, Metcalfe A, Tough S, Kramer MR. International comparison 679 of common risk factors of preterm birth between the U.S. and Canada, using PRAMS 680 and MES (2005-2006). Matern Child Health J 2015; 19:811-8. 681

31. Stocks J, Hislop A, Sonnappa S. Early lung development: lifelong effect on respiratory 682 health and disease. Lancet Respir Med 2013; 1:728-42. 683

32. Thunqvist P, Gustafsson PM, Schultz ES, Bellander T, Berggren-Brostrom E, Norman 684 M, et al. Lung Function at 8 and 16 Years After Moderate-to-Late Preterm Birth: A 685 Prospective Cohort Study. Pediatrics 2016. 686

33. Harju M, Keski-Nisula L, Georgiadis L, Raisanen S, Gissler M, Heinonen S. The burden 687 of childhood asthma and late preterm and early term births. J Pediatr 2014; 688 164:295-9 e1. 689

34. Burke H, Leonardi-Bee J, Hashim A, Pine-Abata H, Chen Y, Cook DG, et al. Prenatal 690 and passive smoke exposure and incidence of asthma and wheeze: systematic 691 review and meta-analysis. Pediatrics 2012; 129:735-44. 692

35. U.S. Department of Health and Human Services CfDCaP, National Center for Health 693 Statistics Smoking Prevalence and Cessation Before and During Pregnancy: Data 694

31

31

From the Birth Certificate, 2014. National Vital Statistics Reports 2016; Vol. 65, No. 695 1:http://www.cdc.gov/nchs/data/nvsr/nvsr65/nvsr_01.pdf. 696

36. Program BCPH. Perinatal Health Report 2008. Provincial Health Services Authority 697 2010:http://www.perinatalservicesbc.ca/Documents/Data-698 Surveillance/Reports/SurveillanceAnnualReport2008.pdf. 699

37. Kirkland G, & Devichand. Women’s Health Surveillance Report: A Multidimensional 700 Look at the Health of Canadian Women. Canadian Institute for Health Information 701 2003; Chapter 6, Gender Differences in Smoking and Self-reported Indictors of 702 Health: 3 and 11. 703

38. Salihu HM, Aliyu MH, Pierre-Louis BJ, Alexander GR. Levels of excess infant deaths 704 attributable to maternal smoking during pregnancy in the United States. Matern 705 Child Health J 2003; 7:219-27. 706

39. Wharry S. Canada a country of two solitudes when smoking rates among 707 anglophones, francophones compared. CMAJ 1997; 156:244-5. 708

40. Al-Sahab B, Saqib M, Hauser G, Tamim H. Prevalence of smoking during pregnancy 709 and associated risk factors among Canadian women: a national survey. BMC 710 Pregnancy Childbirth 2010; 10:24. 711

41. Tong VT, Dietz PM, Morrow B, D'Angelo DV, Farr SL, Rockhill KM, et al. Trends in 712 smoking before, during, and after pregnancy--Pregnancy Risk Assessment 713 Monitoring System, United States, 40 sites, 2000-2010. MMWR Surveill Summ 2013; 714 62:1-19. 715

42. Oyana TJ, Matthews-Juarez P, Cormier SA, Xu X, Juarez PD. Using an External 716 Exposome Framework to Examine Pregnancy-Related Morbidities and Mortalities: 717 Implications for Health Disparities Research. Int J Environ Res Public Health 2016; 718 13:ijerph13010013. 719

43. Azad MB, Konya T, Persaud RR, Guttman DS, Chari RS, Field CJ, et al. Impact of 720 maternal intrapartum antibiotics, method of birth and breastfeeding on gut 721 microbiota during the first year of life: a prospective cohort study. BJOG 2016; 722 123:983-93. 723

44. Cesare Marincola F, Dessi A, Corbu S, Reali A, Fanos V. Clinical impact of human 724 breast milk metabolomics. Clin Chim Acta 2015; 451:103-6. 725

45. Verduci E, Banderali G, Barberi S, Radaelli G, Lops A, Betti F, et al. Epigenetic effects 726 of human breast milk. Nutrients 2014; 6:1711-24. 727

46. Azad MB, Konya T, Maughan H, Guttman DS, Field CJ, Sears MR, et al. Infant gut 728 microbiota and the hygiene hypothesis of allergic disease: impact of household pets 729 and siblings on microbiota composition and diversity. Allergy Asthma Clin Immunol 730 2013; 9:15. 731

47. Laursen MF, Zachariassen G, Bahl MI, Bergstrom A, Host A, Michaelsen KF, et al. 732 Having older siblings is associated with gut microbiota development during early 733 childhood. BMC Microbiol 2015; 15:154. 734

48. Initial Report on Public Health; Breastfeeding Duration. Ministry of Health and Long 735 Term Care 2008; Canadian Community Health Survey 2007:QUEEN'S PRINTER FOR 736 ONTARIO 2008 737

49. Dell S, To T. Breastfeeding and asthma in young children: findings from a 738 population-based study. Arch Pediatr Adolesc Med 2001; 155:1261-5. 739

32

32

50. Sears MR, Greene JM, Willan AR, Taylor DR, Flannery EM, Cowan JO, et al. Long-term 740 relation between breastfeeding and development of atopy and asthma in children 741 and young adults: a longitudinal study. Lancet 2002; 360:901-7. 742

51. Kull I, Almqvist C, Lilja G, Pershagen G, Wickman M. Breast-feeding reduces the risk 743 of asthma during the first 4 years of life. J Allergy Clin Immunol 2004; 114:755-60. 744

52. Victora CG, Bahl R, Barros AJ, Franca GV, Horton S, Krasevec J, et al. Breastfeeding in 745 the 21st century: epidemiology, mechanisms, and lifelong effect. Lancet 2016; 746 387:475-90. 747

53. Maheswaran D, Zeng Y, Chan-Yeung M, Scott J, Osornio-Vargas A, Becker AB, et al. 748 Exposure to Beta-(1,3)-D-glucan in house dust at age 7-10 is associated with airway 749 hyperresponsiveness and atopic asthma by age 11-14. PLoS One 2014; 9:e98878. 750

54. Tischer C, Chen CM, Heinrich J. Association between domestic mould and mould 751 components, and asthma and allergy in children: a systematic review. Eur Respir J 752 2011; 38:812-24. 753

55. Novak M, Vetvicka V. Beta-glucans, history, and the present: immunomodulatory 754 aspects and mechanisms of action. J Immunotoxicol 2008; 5:47-57. 755

56. Dodson RE, Nishioka M, Standley LJ, Perovich LJ, Brody JG, Rudel RA. Endocrine 756 disruptors and asthma-associated chemicals in consumer products. Environ Health 757 Perspect 2012; 120:935-43. 758

57. North ML, Takaro TK, Diamond ML, Ellis AK. Effects of phthalates on the 759 development and expression of allergic disease and asthma. Ann Allergy Asthma 760 Immunol 2014; 112:496-502. 761

58. Hofhuis W, de Jongste JC, Merkus PJ. Adverse health effects of prenatal and postnatal 762 tobacco smoke exposure on children. Arch Dis Child 2003; 88:1086-90. 763

59. Kwon JT, Lee M, Seo GB, Kim HM, Shim I, Lee DH, et al. Cytotoxic effects of air 764 freshener biocides in lung epithelial cells. Nat Prod Commun 2013; 8:1301-4. 765

60. Heinrich J, Gehring U, Cyrys J, Brauer M, Hoek G, Fischer P, et al. Exposure to traffic 766 related air pollutants: self reported traffic intensity versus GIS modelled exposure. 767 Occup Environ Med 2005; 62:517-23. 768

61. Sbihi H, Allen RW, Becker A, Brook JR, Mandhane P, Scott JA, et al. Perinatal 769 Exposure to Traffic-Related Air Pollution and Atopy at 1 Year of Age in a Multi-770 Center Canadian Birth Cohort Study. Environ Health Perspect 2015. 771

62. van de Loo KF, van Gelder MM, Roukema J, Roeleveld N, Merkus PJ, Verhaak CM. 772 Prenatal maternal psychological stress and childhood asthma and wheezing: a meta-773 analysis. Eur Respir J 2016; 47:133-46. 774

775