Citation
Environmental Determinants of Oral Clefts

Material Information

Title:
Environmental Determinants of Oral Clefts
Creator:
Ha, Sandie U
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (153 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Epidemiology
Committee Chair:
XU,XIAOHUI
Committee Co-Chair:
YAGHJYAN,LUSINE
Committee Members:
PEOPLES-SHEPS,MARY D
KAIRALLA,JOHN ANDREW
Graduation Date:
8/8/2015

Subjects

Subjects / Keywords:
Air pollutants ( jstor )
Air pollution ( jstor )
Cleft palate ( jstor )
Disease risks ( jstor )
Infants ( jstor )
Neonatal disorders ( jstor )
Particulate materials ( jstor )
Predisposing factors ( jstor )
Pregnancy ( jstor )
Socioeconomic status ( jstor )
Epidemiology -- Dissertations, Academic -- UF
air-pollution -- birth-defects -- environment -- oral-clefts
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Epidemiology thesis, Ph.D.

Notes

Abstract:
Oral clefts (OC) are the most prevalent forms of non-chromosomal birth defects around the world. Although often not fatal, OC are associated with high morbidity and healthcare expenditure. The etiologies of OC remain largely unknown. This dissertation seeks to contribute to the literature by exploring potential determinants of OC in the social and chemical environment. There is a recent increase in studies on the associations of OC with criteria air pollutants including carbon monoxide (CO), nitrogen dioxide (NO2), ozone (O3), particulate matter with diameter <10 or 5 microns (PM2.5 or PM10), and sulfur dioxide (SO2). A meta-analysis was performed to summarize current evidence on these associations. Results suggest a positive association between SO2 exposure during weeks 3-8 of gestation and cleft lip with or without palate (CLP), and an inverse association between CO exposures during first trimester and cleft palate only (CPO). In Florida, after adjusting for individual risk factors, two significant OC clusters were detected in the central Panhandle region and Polk County, suggesting that there were potential environmental influences. Further analyses showed that neighborhoods, defined by census block groups, with high average annual concentration of PM2.5 or O3, or located in rural areas were positively associated with being part of an OC cluster. No significant associations between neighborhood socioeconomic status and OC cluster were detected. Next, a case-control study was designed to confirm the ecologic associations between air pollution and OC. Results suggested that prenatal exposures to PM2.5 or O3 were positively associated with OC among singleton live births in Florida from 2004- 2008. These associations varied by specific types of OC, time of exposure during pregnancy, maternal smoking status, and infant sex. There was a positive association between PM2.5 exposure during weeks 3-8 of gestation and CLP among infants of mother who smoked during pregnancy. Meanwhile, O3 exposures during both the preconception period and first trimester were associated with both CLP and CPO. The associations with CLP were stronger among male infants; however, the associations with CPO were more pronounced among female infants. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2015.
Local:
Adviser: XU,XIAOHUI.
Local:
Co-adviser: YAGHJYAN,LUSINE.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-02-29
Statement of Responsibility:
by Sandie U Ha.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
2/29/2016
Classification:
LD1780 2015 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

ENVIRONMENTAL DETERMINANTS OF ORAL CLEFTS By SANDIE UYEN HA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2015

PAGE 2

© 2015 Sandie Uyen Ha

PAGE 3

To my parents , Tung Ha and Hanh Le, and younger brother , Sang Ha

PAGE 4

4 ACKNOWLEDGMENTS This dissertation is a product of many years of training , and a network of educators , colleagues, friends , and family who have supported, guided, and inspired me through graduate school. I am writing this acknowledgement on M family. We came to the United States in 1997 empty handed with no understanding of English . My parents, Tung Ha and Hanh Le, worked hard to give my brother and me the opportunity to be the first generation in our family to attend college. I sincerely thank them for their sacrifices, and the patience , hard work, determination, and perseverance that they have taught me all of which have significantly contributed to my accomplishments . Their trust, unconditional love , and support have played the most important part in my career . In addition, I would also like to extend my deepest gratitude to my younger brother, Sang Ha, whose support and encouragement have helped me through difficult time s . I especially thank him for taking care of the family in Washington so I can pursue my academic goal s from far away. I am hugely indebted to my advisor , Dr. Xiaohui Xu, who has been my res earch mentor since I was in the Masters of Public He alth program. This dissertation would not ha ve been possible without his support and guidance. He has taught me how to be a strong and productive researcher. His patience, flexibility , and acceptance ha ve been invaluable in my graduate career. I will always be grateful for the inspiration, encourage ment, and opportunities he has given me over the years. I would also like to thank my supervisory committee members : Dr. John Kairalla, Dr. Mary Peoples Sheps, and Dr. Lusine Yaghjyan . They have given me guidance, encouragement, thoughtful criticisms, time , and attent ion despite their busy schedule. I

PAGE 5

5 also appreciate my former committee member , Dr. Maria Khan , who ha s also given me thoughtful comments and suggestions. Furthermore, I would like to express my gratitude to Dr. Robert Cook and Dr. Jeffrey Roth, two of the many that have been instrumental to my graduate training. Dr. Cook has served as my acade mic advisor over the years . Dr. Roth was my public health internship preceptor, who l ater collaborated with me on many research projects. Their guidance , s upport, and thoughtful feedback have helped me become a stronger scientist and writer. I also thank the Florida Department of Health Off ice of Vital Statistics for giving me access to birth certificate data. I am also grateful for the faculty, staff, and t he wonderful fellow students in the Department of Epidemiology at the University of Florida . I specifically want to thank my research team mate , Hui Hu, who has generously helped me on many occasions with data issues. Last, but not least, I am especially t han kful for my significant other, D ung Nguyen, for his support, understanding , and the editing of this dissertation.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 10 ABSTRACT ................................ ................................ ................................ ................... 13 CHAPTER 1 BACKGROUND ................................ ................................ ................................ ...... 15 Public Health Significance of Oral Clefts ................................ ................................ . 15 Theoretical Framework ................................ ................................ ........................... 17 Risk F actors of Oral Clefts ................................ ................................ ...................... 19 Individual Risk Factors ................................ ................................ ..................... 19 Environmental Risk Factors ................................ ................................ .............. 22 Biological and physical environment ................................ .......................... 23 Social environment ................................ ................................ .................... 23 Chemical environment ................................ ................................ ............... 24 Air Pollution and Oral Clefts ................................ ................................ .................... 26 Knowledge Gap ................................ ................................ ................................ ...... 27 Research Objec tives ................................ ................................ ............................... 31 2 ASSOCIATIONS BETWEEN AIR POLLUTION AND ORAL CLEFTS: A SYSTEMATIC REVIEW AND META ANALYSIS ................................ .................... 36 Introduction ................................ ................................ ................................ ............. 36 Materials and Methods ................................ ................................ ............................ 38 Search Strategies ................................ ................................ ............................. 38 Meta analysis ................................ ................................ ................................ ... 38 Results ................................ ................................ ................................ .................... 41 Qualitative Results ................................ ................................ ........................... 41 Quantitative Results ................................ ................................ ......................... 43 Discussi on ................................ ................................ ................................ .............. 44 Participant Selection ................................ ................................ ......................... 44 Exposure Assessment ................................ ................................ ...................... 45 Comparability and Confounding ................................ ................................ ....... 47 Biologic Plausibility ................................ ................................ ........................... 48 Limitations ................................ ................................ ................................ ........ 49 Future Directions ................................ ................................ .............................. 50 Conclusion ................................ ................................ ................................ .............. 51

PAGE 7

7 3 LOCALIZED CLUSTERING AND NEIGHBORHOOD LEVEL DETERMINANTS OF ORAL CLEFTS IN FLORIDA ................................ ................................ ............ 65 Introduction ................................ ................................ ................................ ............. 65 Materials and Methods ................................ ................................ ............................ 68 Setting and Participants ................................ ................................ .................... 68 Cluster Analysis ................................ ................................ ................................ 69 Neighborhood Predictors of Oral Cleft Clustering ................................ ............. 71 Results ................................ ................................ ................................ .................... 73 Cluster Analysis ................................ ................................ ................................ 73 Neighborhood Predictors of Oral Cleft Clustering ................................ ............. 74 Discussion ................................ ................................ ................................ .............. 76 Conclusion ................................ ................................ ................................ .............. 79 4 ASSOCIATIONS BETWEEN PRENATAL EXPOSURES TO AIR POLLUTION AND ORAL CLEFTS ................................ ................................ ............................... 87 Background ................................ ................................ ................................ ............. 87 Materials and Meth ods ................................ ................................ ............................ 90 Setting and Participants ................................ ................................ .................... 90 Exposure Assessment ................................ ................................ ...................... 90 Outcome Assessment ................................ ................................ ...................... 91 Covariates ................................ ................................ ................................ ........ 92 Statistical Analysis ................................ ................................ ............................ 92 Results ................................ ................................ ................................ .................... 93 Discussion ................................ ................................ ................................ .............. 96 Conclusion ................................ ................................ ................................ ............ 100 5 CONCLUSIONS ................................ ................................ ................................ ... 112 Summary of Research Objectives ................................ ................................ ........ 112 Accomplishments of this Dissertation ................................ ................................ ... 113 Meta analysis of Relevant Literature ................................ .............................. 113 Identification of Oral Clefts Clusters in Florida ................................ ................ 114 Determination of Neighborhood Predictors of Oral Clefts Clustering .............. 115 Determination of Association of Air Pollution with Oral Clefts and Identification of Susceptible Subgroups ................................ ...................... 116 Biological Mechanisms Linking Air Po llution and Oral Clefts ................................ 117 Oxidative Stress and Inflammation ................................ ................................ . 118 Alteration of Endothelial, Rheologic and Hemodynamic Functions ................ 119 Endocrine Disruption ................................ ................................ ...................... 120 Genetic and Epigenetic Changes ................................ ................................ ... 122 Future Directions ................................ ................................ ................................ .. 122 LIST OF RE FERENCES ................................ ................................ ............................. 130 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 153

PAGE 8

8 LIST OF TABLES Table page 2 1 Characteristics of studies included in the review ................................ ................ 52 2 2 Summary risk estimates of the associations between criteria air pollutants and oral clefts ................................ ................................ ................................ ..... 58 2 3 Sensitivity analyses results for summary risk estimates of the associations between criteria air pollutants and oral clefts ................................ ...................... 59 3 1 Spatial cluster analysis of oral clefts in Florida, 2004 2008 ................................ 81 3 2 Neighborhood characteristics of census block groups in side and outside of oral cleft clusters in Florida, 2004 2008 ................................ .............................. 82 3 3 Associations between neighborhood characteristics and oral c left clusters in Florida, 2004 2008 ................................ ................................ ............................. 8 3 4 1 Characteristics of study participants in Florida, 2004 2008 .............................. 102 4 2 Estimated exposure to ambient air pollutants at birth residence for cases and controls in Florida, 2004 2008. ................................ ................................ ......... 104 4 3 Associations between PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy among Florida births, 2004 2008. ........... 105 4 4 Associations between O 3 and oral clefts by exposure windows and infant sex among Florida births, 2004 2008. ................................ ................................ ..... 106 4 5 Associations between PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy in a multi pollutant model, 2004 2008. .... 107 4 6 Associations between O 3 and oral clefts by exposure windows and infant sex in a multi pollutant model, 2004 2008. ................................ .............................. 108 4 7 Associations between nearest monitor PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy among Florida births, 2004 2008. ................................ ................................ ................................ ....... 109 4 8 Associations between nearest monitor O 3 and oral clefts by exposure windows and infant sex amo ng Florida births, 2004 2008 ................................ 110 5 1 Methods to reduce personal exposure to outdoor air pollution ......................... 127

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Theoretical framework guiding present research.. ................................ .............. 34 1 2 Trend in studies on the association between air pollution and birth defects, 2000 2014 ................................ ................................ ................................ .......... 35 2 1 Selection of relevant studies included in the study ................................ ............. 61 2 2 Forest plot of summary estimate of the association between CO exposure during first trimester and CPO. ................................ ................................ ........... 62 2 3 Forest plot of summary estimate of the association between SO 2 exposure during weeks 3 8 and CLP. ................................ ................................ ................ 63 2 4 Funnel plots for the associations between air pollutants and OC . ...................... 64 3 1 Spatial scanning method using ................................ .... 84 3 2 Spatial clusters of oral clefts in Florida using univariate models , 2004 2008.. .... 85 3 3 Spatial clu sters of oral clefts in Florida using bivariate models, 2004 2008.. ...... 86 4 1 Distribution of active air monitors in Flori da during the study period ................. 111 5 1 Summary of findings ................................ ................................ ......................... 128 5 2 Plausible biological mechanisms linking air pollution and oral clefts ................ 129

PAGE 10

10 LIST OF ABBREVIATIONS ADH1C A lcohol dehydrogenase 1C ART Assisted reproductive therapy BMI Body mass index BMP 4 B one morphogenetic 4 AQS Air quality system CBG Census block group CI Confidence interval CLP Cleft lip with or without cleft palate CMAQ Community multi scale air q uality CO Carbon monoxide CO Hb Carboxyhemoglobin CPO Cleft palate only DNA Deoxyribonucleic acid EPA Environmental Protection Agency ETS Environmental tobacco smoke GIS Geographic information system Hb Hemoglobin HBM Hierarchical Bayesian m odel IQR Interquartile range IRF6 I nterferon regulatory factor 6 IUGR Intrauterine growth restriction LBW Low birth weight LLR Likelihood ratio LMP Last menstrual period

PAGE 11

11 MTHFR M ethylenetetrahydrofolate reductase MW Molecular weight NO x Nitrogen oxides NO 2 Nitrogen dioxide O 3 Ozone OC Oral clefts OR Odds ratio PAH P olycyclic aromatic hydrocarbon Pb Lead PM Particulate matter PM 2.5 Particulate matter with aerodynamic diameter less than 2.5 micrometers PM 10 Particulate matter with aerodynamic diameter less than 10 micrometers p pb Parts per billion p pm Parts per million Pphm Parts per hundred million PTD Preterm delivery PVRL1 P oliovirus receptor like 1 RR Relative risk SC Spontaneous conception SD Standard deviation SES Socioeconomic status S O x Sulfur oxides SO 2 Sulfur dioxide TBX22 T box transcription factor 22

PAGE 12

12 TGFA T ransforming growth factor alpha US United States

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ENVIRONMENTAL D ETERMINANTS OF ORAL CLEFT S By Sandie Uyen Ha August 2015 Chair: Xiaohui Xu Maj or: Epidemiology Oral clefts ( OC) are the most prevalent form s of non chromosomal birth defect s around the world . Although often not fatal , OC are associated with high morbidity and high healthcare expenditure . The etiologies of OC remain largely unknown. This dissertation seeks to contribute to the literature on OC etiologies by exploring potential determinants in the environment . There is a recent increase in studies on the associations of OC with criteria air pollutants that include ca rbon monoxide (CO), nitrogen dioxide (NO 2 ), ozone (O 3 ), particulate matter with diameter < 10 or 5 µm (PM 2.5 or PM 10 ), and sulfur dioxide (SO 2 ). First, a meta analysis was performed to summarize current evidence on th ese association s . Results suggest ed a positive association between SO 2 exposure during weeks 3 8 of gestation and cleft lip with or without palate (CLP) , and an inverse association between CO exposures during first trimester and cleft palate only (CPO) . In the state of Florida f rom 2004 2008, after adjusting for individual risk factors, two significant OC clusters were detected in the P anhandle region and Polk Count y , suggesting that there were potential influences from environmental factors . Further ecologic analyses showed that neighborhoods , defined by census block groups , with

PAGE 14

14 high er average annual concentration s of PM 2.5 or O 3 and/or those located in rural areas were significantly more likely to be part of an OC cluster. No significant associations between neighborhood socioeconomic status and OC cluster ing were detected. Due to ecological fallacy, a population based case control study was designed to confirm the ecologic associations of PM 2.5 and O 3 with OC previously found . Results suggest ed that exposures to PM 2.5 or O 3 during the first trimester of pregnancy were positively associated with OC among singleton live births in Florida from 2004 2008 . T hese associations varied by specific types of OC, tim ing of exposure , maternal smoking status, and infant sex. Specifically, th ere was a positive association between PM 2.5 exposure during weeks 3 8 of gestation and CLP among infants of moth er s who smoked during pregnancy. Meanwhile , O 3 exposures during all exposure windows within the first trimester were positively associated wi th CLP and CPO. For CPO, these associations were more pronounced among female infants compared to male infants .

PAGE 15

15 CHAPTER 1 BACKGROUN D Public Health Significance of Oral Clefts A b irth defect is defined as a condition present at birth that cause s structural or developmental changes in one or more parts of the body . Birth defect s are currently the leading cause of death among infants, accounting for more than 20% of all infant mortality ( Martin et al., 2008 ) . Every year, one out of 33 births is born in the United States (US) with at least one major birth defect. Oral cleft s (O C ) are the most common form s of non chromosomal major birth defect s with a prevalence of 1.7 per 1000 birth s , which is equivalent to approximately 7 , 100 cases annually in the U S ( Parker et al., 2010 ) . There are two main types of OC : cleft lip and cleft palate. These clefts may occur separately or in combination. For etiologic reasons, they are divided into two groups: cleft lip w ith or without cleft palate (CL P) and cleft palate only (CPO). C left lip with or without cleft palate occurs more often and accounts for approximately 63% of all OC ( Parker et al., 2010 ) . Although isolated OC are often not as fatal as other types of birth defect, they are associated with two times higher risk of infant mortality within the first year of life c ompared to the unaffected population (Carlson et al., 2013). In addition, they are associated with high morbidity . I nfants who have OC experience significantly higher risk of severe recurring ear infections as well as feeding, hearing , speech, and dental complications ( Conrad et al., 2014 ; de Vries et al., 2014 ; Sharma and Nanda, 2009 ; Tannure et al., 2012 ) . Most cases of OC require multiple surgical intervention s to minimize developmental complicatio ns later in life. Consequently, infant s affected with OC experience significantly higher cost of healthcare and management compared to those who are unaffected ( Boulet et al., 2009 ; Cassell et al., 2008 ; Wehby and Cassell,

PAGE 16

16 2010 ; Wehby et al., 2012 ; Weiss et al., 2009 ; Yazdy et al., 2008 ) . The mean cost of hospitalizations associated with OC from birth to tw o years of age for each case is approximately $21,090 while that for an unaffected infant is only about $2,504 ( Weiss et al., 2009 ) . Furthermore , OC are also associated with high er risk of long term disabilities ( Moraleda Cibrian et al., 2014 ; Mossey and Modell, 2012 ; Wehby et al., 2014a ; Wehby et al., 2014b ) , decreased psychological functioning, and lower quality of life for b oth the infants and their families ( Antunes et al. , 2014 ; Marcusson et al., 2001 ; Wehby and Cassell, 2010 ) . In 1998, the US has implemented folic acid fortification in the national food supplies t o prevent birth defects. It is also recommended that all women of child bea ring age consume 400µg of folic acid daily to prevent birth defects. Despite evidence of a decline in the rates of birth defect s following this intervention in the US , this decline is observed for primarily neural tube defect s ( CDC, 1992 ; CDC, 1993 ; CDC, 2004 ; Honein et al., 2001 ; Williams et al., 2005 ) . Currently , no consistent evidence exists to support that folic acid fortification has significantly reduced the prevalence of OC ( Hashmi et al ., 2005 ; Simmons et al., 2004 ) . Therefore , most OC cases are likely caused by other factors . Today, there is no known effective intervention to prevent OC . On top of that, there is still evidence of significant disparities in the prevalence of OC with respect to d emographic and geographic characteristics ( Canfield et al., 2006a ; Poletta et al., 2007 ) . Consequently, further understanding of the eti ology and directing efforts towards preventi on of OC remain important priorities in maternal and child health management and research as highlighted in Healthy People 2020 objectives ( Healthy People 2020, 2014 ) .

PAGE 17

17 Theoretical F ramework F actors that influence the risk of OC can be explained usi ng the socioecological framework , which postulates that a health outcome is influenced by the product of numerous individual and environmental factors ( Bronfrenbrenner, 1979 ; May, 1958 ) . These factors may also interact with each other to influence health outcomes throughout the life course of individuals, families, and communities. In this context, the risk factors of a disease or disorder c an be organized into a theoretical framework consisting of two main categories: individual (of both the mother and infant) and environment al risk factors. In dividual risk factors are comprised of characteristics at the individual leve l that are not part of the exogenous environment. These factors may include genetic s , individual biology, and other innate characteristics . Environmental risk factors may include factors from the environment that are external to a person . They can be furth er divided into four subcategories: the biological environment, the physical environment, the social environment, and the chemical environment. The biological environment is composed of living organisms such as pathogens . The physical environment is composed of physical attributes of the environment such as weather, radiation, heat, noise, etc. T he socia l environment includes the socioeconomic context in which individuals live . The chemical environment is comprised of chemical substances such as expos ures to drugs or hazardous pollutants. From an intervention perspective, environmental risk factors are more important because they can be amendable either at the individual level (e.g. change in behavior) or the societal level ( e.g. policy or regulation) . Thus, to support intervention efforts, this dissertation focuses on exploring the environmental risk factors for OC after adjusting for known individual risk factors. In addition, while there is a lack of evidence on the

PAGE 18

18 association of the biological and physical environment with OC, individual risk factors, the social environment, and the chemical environment have been implicated in the etiology of OC ( Mossey et al., 2009 ) . Therefore, the framework guiding the current res earch emph asizes risk factors in the individual level, the social, and the chemical environment. The risk factors within these specific categories are reviewed in more details in the next section. Figure 1 1 illustrates the theoretical framework for the determinants of OC. In addition to their independent effects , individual and environmental risk factors in the theoretical framework may also interact with each other to influence the risk of OC in several ways : 1 ) differential high risk exposures across the population ; 2 ) differential susceptibility to OC across the population ; or 3 ) a combination of both differential high risk exposures and susceptibility to OC. The first interaction involves unequal distribution of high risk exposures such as hazardous chemical s among certain sub group s of the population , leading to higher risk of OC . The second interaction may suggest higher susceptibility to OC in certain subgroups of the population even at the same level of hazardous exposure , particu larly the more disadvantaged . Many authors have hypothesized that more socioeconomically disadvantaged groups may be more sensitive to environmental exposures due to their underlying higher burden of existing health conditions ( Rios et al., 1993 ; Sexton, 1997 ; Sexton et al., 1993 ) . The third interaction s uggests that some subgroups may be simultaneously exposed to higher load of environmental exposures and have increased underlying susceptibility to OC. Based on this theoretical framework, more studies are needed to contribute to the fur ther understanding OC etiology for effective intervention by identifying : 1)

PAGE 19

19 environmental factors tha t may increase the risk of OC ; 2 ) potential cluster(s) of the population who may be more at risk for OC even after adjusting for individual risk factors ; and 3) subgroups that are more sensitive when exposed to these environmental risk factors . First, the identification of environmen tal risk factors for OC allows intervention to reverse or minimize the exposure. Second , ident ification of high risk clustering after adjust ing for individual risk factors may provide insights into potential enviro nmental contributors . It also provide s critical information for allocation of resources and services. Third , the identification of subgroups that are more sensitive when exposed to environmental risk factors provide s useful information for targeted intervention . It also aid s with the understanding of the underlying mechanisms of the association between environmental risk factor s and OC . Risk Factors of Oral Clefts The risk factors of OC remain largely unknown for most cases. However, population based studies have suggested that they may be multifaceted ( Mossey et al., 2009 ) . These risk factors can be organized based on the theoretical framework previously proposed and are briefly discussed below. Individual Risk Factors Individual r isk factors have received relatively more attention in the literature compared to environmental risk factors. They include innate maternal and fetal characteristics including genetic s , maternal comorbidities, and demographics . S everal genes have been identified that are associated with the risk of the development of OC among affected families around the world ( Kohli and Kohli, 2012 ) . These genes include the T box transcription factor 22 (TBX22), poliovirus receptor like 1 (PVRL1), and the interferon regulatory factor 6 (IRF6) gene ( Braybrook et al., 2001 ; Kondo et al., 2002 ;

PAGE 20

20 Sozen et al., 2001 ; Suzuki et al., 2000 ) . In a recent meta analysis of 20 studies , poly morphism in the transforming growth factor a lpha (TGFA) gene was also found to be associated with over 50% increased risk of OC ( Feng et al., 2014 ) . Other genetic polymorphisms that have be en implicated in the etiology of OC include the Msh homeobox homolog 1 (MSX) , me thylenetetrahydrofolate reductase ( MTHFR ) ( Jagomagi et al., 2010 ; Kohelet et al., 1990 ; Souza et al., 2013 ) , and the bone morphogenetic (BMP 4) gene ( Antunes et al., 2013 ) . Despite the important role of genetics , it accounts for less than 20% of OC occurrences ( Grosen et al., 20 11 ) . In addition to genetic s , maternal demograph ics have also been r ecognized as risk factors for OC an d other ty pes of birth defect s . Specifically , very young or very old maternal age has been linked to higher risk of having infants with OC in a number of studies ( Bille et al., 2005 ; DeRoo et al., 2003 ) . For example, in a la rge California study, women aged 39 years or older were twice as likely to have a child with OC compared to women age d between 25 29 years ( Shaw et al., 1991 ) . Other studies also showed that maternal age under 20 is significantly associated with higher risk of having an infant affect ed by OC ( DeRoo et al., 2003 ; Reefhuis and Honein, 2004 ) . O n the contrary, some studies also failed to show that maternal age influences risk of OC ( Abramowicz et al., 2003 ; Baird et al., 1991 ; Baird et al., 1994 ; Vieira et al., 2002 ) . A lthough it is generally accepted that maternal age is a risk factor, the evidence supporting this association is still inconsistent. Maternal race is also a major risk factor for OC in the US. The prevalence of OC is lower among infants of Black and Hispanic women (0.62 per 1,000 for both ) , but higher among Native American s (1.99 per 1,000) compared to Whites (1.05 per 1,000)

PAGE 21

21 ( Croen et al., 1998 ; Forrester and Merz, 2004 ; Lebby et al., 2010 ) . Among Asians and Pacific Islanders , the prevalence is higher among some far East Asian groups such as Vietnamese (1.23 per 1,000) and Laotian (1.46 per 1,000) compared to Pacific Islanders (0.83 per 1,000) ( Croen et al., 1998 ; Forrester an d Merz, 2004 ) . The reasons for racial differences in risk are currently not clear. Like many other health outcomes, m aternal socioeconomic status (SES) may also play a role ; however, the evidence is still limited and inconsistent . Some s tudies have suggested that babies born to mother with lower SES indicated by lower education or lower income have higher risk of having birth defects ( Acuna Gonzalez et al., 2011 ; Damiano et al., 2009 ; Dvivedi and Dvivedi, 2012 ; Ma et al., 2015 ; Marshall et al., 2010 ; Schembari et al., 2014 ; Yu et al., 2014 ) . Nevertheless, a firm association between individual SES and OC has not been established in the US ( Mossey et al., 2009 ) . Some contributing reasons may include d ifferences in the measurement and classifi cation of SES. M aternal comorbidities may also contribute to the risk of OC . In a large multicenter study of mothers of infants who were born with and without birth defects , mothers with gestational diabetes ha d approximately 50% in crease in the risk of having infants with OC ( Correa et al., 2008 ) . Maternal o besity kg/m 2 ) has also been shown to increase the risk of having an infant affected by OC by approximately 26 % compared to normal weight (BMI <25 kg/m 2 ) ( Molina Solana et al., 2013 ) . A recent meta analysis of 12 studies has also suggested that obese and overweight women have a 16% and 14% inc reased risk of having an infant with CLP, respectively ( Izedonmwen et al., 2015 ) . Although not well established, maternal infection during pregnancy has

PAGE 22

22 also been associated the risk of birth defects in general , but not for OC specifically ( Dreier et al., 2014 ; Shahrukh Hashmi et al., 2010 ) . This association is difficult to assess because OC develop very early in pregnancy (weeks 3 8 of gestation ) while maternal infection can occur at various time during pregnancy . Lastly , male infants are more disproportion ately affected than females ( Blanco Davila, 2003 ) . When stratified by specific phenotypes of OC, male infants are more likely to have CL P while female infants are at a higher risk for CPO ( Blanco Davila, 2003 ; Das et al., 1995 ) . The reasons underlying the different risk s between male and female infants are currently unknown. However, it has been suggested that maternal hormonal profile around the time of conception may partially influence the sexes of the offspring ( James, 2004 ; James, 2008a ) . Therefore, some authors have hypothesized that the ho rmone profiles that control fetal sex may be related to the mechanisms that cause OC, one of which may involve endocrine disruption ( James, 2000 ; Kannan et al., 2006 ; Shkoukani et al., 2013 ) . Environmental Risk Factors As previously discussed, the environment may be categorized into four types: the biological environment, the physical environment, the social environment, and the chemical environment. There is a lack of evidence linking the biological or physical environm ent with OC. However, the social environment and the chemical environment have been suggested to have potential influences on OC ( Leite et al., 2002 ; Mossey et al., 2009 ) . Therefore, these latter two categorie s of environmental factors are emphasized in the theoretical framework guiding this dissertation .

PAGE 23

23 Biological and physical environment Currently, there is in sufficient evidence to support the potential influence of the biological or the physical environment on the risk of OC. As previously mentioned, although maternal infection during pregnan cy has been associated with a higher risk of having an infant with birth defects in some instances, no pathogens have been identified to increase the risk of OC ( Dreier et al., 2014 ; Shahrukh Hashmi et al., 2010 ) . These associations, if they truly exist, are likely due to the physiologic changes occurring during the infection, not the infec tious agent itself since no specificity has been established. Similarly, there is a lack of evidence for potential association between the physical environment and risk of OC. Social environment In terms of the social environment , t here is a relatively n ew body of research , that emphasizes the importance of contextual determinants of disease outcomes ( Diez Roux, 2001 ) . Specifically, while the individual characteristics previously described are important, a disease outcome could also be influenced independently by the neighborhood context in which the individual lives . The understanding of neighborhood level social determinants of OC is important because it may provide etiologic clues, help improve our understanding of SES as a potential confounder of other exposures, and help in health services plannin g as well as intervention efforts. For these reasons , it is important to identify neighborhoods that are at an increased risk and explore the neighborhood characteristics that may be associated with high OC rates . Therefore, a few studies around the world have investigated the potential impact of neighborhood level characteristics on the risk of OC ( Carmichael et al., 2009 ; Clark et al., 2003 ; Durning et al., 2007 ; Rodrigues et al., 2009 ) .

PAGE 24

24 However, evidence is still limited and inconsiste nt . More specifically, l ow neighborhood SES has been shown to be a risk factor for OC in some studies ( Clark et al., 2003 ; Durning et al., 2007 ) but not others ( Carmichael et al., 2009 ; Rodrigues et al., 2009 ; Vrijheid et al., 2000 ) . Thus, m ore studies are still needed to further investigate whether neighborhood SES may influence the risk of OC . Chemical e nvironment Th e chemical environment may include any chemical substance or compound that pregnant women may be exposed to. There is eviden ce showing that exposures to folic acid, tobacco smoke, and some other substance s may impact the risk of OC ( Botto et al., 2014 ; Carmichael et al., 2007 ; Chung et al., 2000 ; Cornel et al., 1996 ; Goh et al., 2006 ; Honein et al., 2007 ) . Although the association between folic acid intake and OC is not as strong as that for neural tube defects, low levels of folic acid in take during periconception time have been associated with a higher risk of OC ( Figueiredo et al., 2015 ) . On the contrary, some studies found a positive association where higher folic acid concentration was associated with a higher risk ( Rozendaal et al., 2013 ) , while other s found no association with OC ( Little et al., 2008 ; Shaw et al., 2006 ) . The inconsistent findings may be attributed to exposure misclassification as the result of self reported information . Overall, despite some evidence that fol ic acid has a protective effect against birth defects , specifically neural tube defects , the findings are still inconsistent for OC ( Wehby and Murray, 2010 ) . Maternal smoking during pregnancy is one of the most established risk factors for OC . An e arlier meta analysis including 11 studies ha s shown that maternal smoking during the first trimester incr ease d t he risk of CL P by 29% and CPO by 32% ( Wyszynski et al., 1997 ) . More recent meta analyses also suggested higher risk among women who

PAGE 25

25 smoke d during pregnancy with 34% increase for CL P , 22% for CPO ( Little et al., 2004 ) , and 48% for both outcomes combined ( Molina Solana et al., 2013 ) . H owever, there is not enough evidence to suggest an association between OC and passive smoking ( Honein et al., 2007 ) . Alcohol consumption during pregnancy, especially during first trimester , is also associated with OC. In a population based case control study taken place in Norway from 1996 2001, mothers who consumed five or more alcoholi c drinks per sitting during the first trimester of pregnancy had a 2.6 fold increase d risk of having an affected infant ( Boyles et al., 2010 ) . This association was stronger among women with a mutation in the alcohol dehydrogenase 1C (ADH1C) gene , which is needed to properly metabolize alcohol . Similar findings were also observed in another large study ( Grewal et al., 2008 ) as well as a recent meta analysis ( Molina Solana et al., 2013 ) . Meanwhile, low level s of alcohol consumption during pregnancy (less than 3 drinks a week) have not been associ ated with OC ( Meyer et al., 2003 ) . E xposure s to other substances including illicit drugs such as methamphetamine, cocaine, and marijuana ( Forrester and Merz, 2007 ) ; and some prescription drugs including steroids ( Kohli and Kohli, 2012 ) and anticonvulsant s used to treat epilepsy or migraine headaches , may also increase the risk of having an OC affect ed infan t ( Alsaad et al., 2015 ; Holmes et al., 2008 ; Holmes and Hernandez Diaz, 2012 ) . The potenti al association between hazardous chemical s in the broader environment and OC has been gaining increasing attention by researchers in recent years . For example, m aternal occupational e xposures to certain organic substances including polycyclic aromatic hydrocarbon (PAH) ( Langlois et al., 2013 ) and organic

PAGE 26

26 solvents have been found to be associated with OC ( Chevrier et al., 2006 ; Leite et al., 2002 ) . However, these exposures have not been extensively studied. Of particular not e , due to its omnipresent nature , air pollution has been getting more attention in recent year s for its potent ial association with birth outcomes including low birthweight, preterm delivery, growth restriction, birth defects, and OC ( Bonzini et al., 2010 ; Vrijheid et al., 2011 ) . Despite the increase in the number of studies , it is currently difficult to determine whether an overall association ex ists between air pollution and OC due to the differences in study design . More details on this topic are discussed in the next section. Air Pollution and Oral C left s Despite concerted public health efforts towards environmental control in the past decades, a significant portion of the US population is still living in ar eas with sub optimal levels of air quality . Air pollution consists of a diverse mixture of gases and particulate matter that includes organic and inorganic compounds comi ng from various sources , including traffic and industrial processes. Currently, the Environmental Protection Agency ( EPA ) regulates six common air pollutants, also called criteria air pollutants. They include particulate matter (PM) with aerodynamic diamet er less than 2.5 or 10 m (PM 2.5 or PM 10 ), ground level ozone (O 3 ) , carbon monoxide (CO) , sulfur oxides (SO x ) , nitrogen oxides (NO x ) , and lead (Pb) . These pollutants can harm human health , the environment, and cause property damage. According to the EPA , PM 2.5 and O 3 are the air pollutants that currently have the most widespread health threat . There i s mounting evidence linking PM 2.5 and O 3 to many serious health endpoints. These endpoints include mortality ( Hoek et al., 2013 ; Schwartz, 1994 ; Shang et al., 2013 ; Yan et al., 2013 ) , respiratory diseases ( Brunekreef et al., 2009 ; Gent et al., 2003 ; Kelly and Fuss ell, 2011 ; Laumbach and Kipen, 2012 ) , cardiovascular diseases ( Gill et al., 2011 ;

PAGE 27

27 Su e t al., 2011b ; Uzoigwe et al., 2013 ; Xu et al., 2013 ; Yamamoto et al., 2014 ) , cancer ( Demetriou et al., 2012 ; Raaschou Nielsen and Reynolds, 2006 ) and many other health outcomes ( Cho et al., 2014 ) . Meanwhile, a ssociation s between exposure to PM 2.5 or O 3 and the risk o f adverse birth outcomes such as infant mortality ( Loomis et al., 1999 ; Ritz et al., 2006 ; Woodruff et al., 1997 ) , low birth weight ( Bobak, 2000 ; Romao et al., 2013 ) , preterm delivery ( Bobak, 2000 ) , and intrauterine growth restriction ( Pereira et al., 2012 ) have also been suggested . Due to the posit ive association between air pollution and adverse birth outcomes, it is plausible to hypothesize that exposure to air pollutants such as PM 2.5 and O 3 may also be associa te d with the risk of OC and birth defects . A quick literature search in Pubmed using the keywords air pollution and or congenital anomalies suggests that the number of studie s investigating the association between air pollution and birth defect s has increased in recent years (Figure 1 2 ). Th ere has also been a recent increase in the number of studies linking criteria air pollutants and OC ( Farhi et al., 2014 ; Padula et al., 2013b ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ; Vinikoor Imler et al., 2015 ) . However, g iven differences in study designs across studies, it is still difficult to draw a fir m conclusion on the association between criteria air po llutants and OC. Knowledge G ap As suggested by the theoretical framework, in order to further understand the etiolog y of OC for development of effective interventions, more studies are needed to identify environmental risk factors (e.g. social and chemical) , high risk subgroups (e.g. clusters) , and subgroups that may be more sensitive to the effects of hazardous environmental exposures on OC risk .

PAGE 28

28 As discussed in the previous section , large heterogeneity in design s among increasin g number of studies has made it difficult to conclude whether an overall association between air pollution and OC exists . Moreover, e xisting literature on air pollution and birth defects shows mixed evidence with some studies suggesting positive associatio n ( Cordier et al., 2010 ; Dadvand et al., 2011a ; Dadvand et al., 2011b ; Dolk et al., 2010 ; Gilboa et al., 2005 ; Hansen et al., 2009 ; Lupo et al., 2011 ; Ritz et al., 2002 ; Strickland et al., 2009 ) , mixed association ( Cordier et al., 2004 ; Hwang and Jaakkola, 2008 ; Langlois et al., 2009a ; Rankin et al., 2009 ) , and no association ( Marshall et al., 2010 ; Queisser Luft et al., 2011 ; Vrijheid et al., 2002 ) . Prior studies have provided enough evidence suggesting positive associations between air pollution and congenital heart defects ( Chen et al., 2014 ) . However, the association for OC is still unclear. Therefore , the re is a strong need for a systematic review and meta analysis of the literatur e pertaining to the association between criteria air pollutants and OC in order to determine whether there is an overall association. To tie this back to the theoretical framework, this type of study will help to determine whether exposure to air pollution is a risk factor for OC , and the findings may help to guide future studies. Furthermore, a s previously mentioned, PM 2.5 and O 3 are cu rrently the two pollutants with the most widespread health implications. Despite significant associations with many adverse birth outcomes, the associations of these two pollutants with OC are not clear ( Proietti et al., 2013 ) , suggesting the need for more studies . For OC, the risk factors previously discussed may be unevenly distributed across geographic areas. Thus, spatial variation in the rate of OC may also exist. Therefor e, investigation of the spatial patterns of OC can be of great public health value .

PAGE 29

29 Identification of spatial clusters with unexpectedly high number of cases of OC at small area level s may reveal potential determinants such as hazardous environment al exposures or other neighborhood level determinants. Furthermore, identification of high risk areas is also important for health service planning, allocation of resources, and the understanding of environmental etiology . O ral clefts are particularly well s uited for this type of spatial analyses because the lag time between exposure to environmental and high risk social and chemical conditions and the development of a birth def ect outcome is relatively short . However, to our knowledge, no studies have been c onducted to identify potential OC high risk areas in the state of Florida or even in the US. Florida is am ong the states with a large number of annual births (~240,000). Furthermore, its population diversity makes Florida an ideal setting to study such phe nomen on . Additionally, researchers have long recognized that living in certain areas has an independent impact on health outcomes. For this reason, neighborhood level risk factors have been identified for many health outcomes including cardiovascular disea se ( Hamano et al., 2013 ) , cancer ( Bethea et a l., 2014 ) , mental health ( Fone et al., 2013 ) , and adverse birth outcomes ( Agyemang et al., 2009 ; Metcalfe et al., 2011 ) . However, the answer to this questi on with respect to OC and other birth defects is less explored. As discussed in the social environme nt section, the few existing studies on this topic observed inconsistent associations ( Carmichael et al., 2003 ; Clark et al., 2003 ; Durning et al., 2007 ; Rodrigues et al., 2009 ; Vrijheid et al., 2000 ) . Thus, th ere is still a need for more studies to explore neighborhood risk factors for OC

PAGE 30

30 Furthermore , cons istent with the theoretical framework previously described, air pollution studies have suggested that different subgroups of the population , defined by gender, race, lifestyle, etc., may be affected differently by air pollution ( Dubowsky et al., 2006 ; Zanobetti et al., 2000 ; Zeka et al., 2006 ) . These studies suggested that air pollution studies should identify susceptible groups to make better risk prediction . Meanwhile , characteristics such as infant sex, maternal race, and maternal lifestyle factors including smoking during pregnancy have been consistently shown as strong risk factors for OC ( Blencowe et al., 2010 ; Hackshaw et al., 2011 ; Jia et al., 2011 ; Kan et al., 2008 ; Mateja et al., 2012 ; Zanobetti and Schwartz, 2000 ) . T herefore, it is plausible to hyp othesize that the association between air pollution and OC , if it exists, may be stronger among subgroups of the population who are already at a higher risk d ue to these risk factors (e.g. effect modifiers) . For example, t here is evidence supporting that air pollution and exposure to smoking may have synergic effects on health ( Turner et al., 2014 ) . In addition, studies on the association between air pollution and adverse birth outcomes (e.g. low birth weight and preterm delivery) have also suggested that these associations may differ by infant sex ( Ghosh et al., 2007 ) . Other authors have also reported that the health effects of air pollution may also differ by race/ethnicity ( American Lung A ssociation, 2001 ; Sexton et al., 1993 ) . However, to our knowledge no studies have examined these potential effect modifiers on the association between air p ollution and OC . The National Research Council has identified susceptible population s as a key knowledge g ap in studies of health effects of air pollution, and has emphasized the need for future stud ies to identify them ( National Research Council, 2004 ) . Identification

PAGE 31

31 of such populations may help with the further understanding of speci fic biological mechanisms underlying the observed association between air pollution and health outcomes. In addition, a better knowledge of these subgroups will be helpful for developing effective interventions, public policy , and risk assessments. Resear ch O bjectives Given the lack of evidence linking air pollution as well as neighborhood level characteristics to OC risk , and the lack of literature on susceptible subgroups, it is timely and important to address these knowledge gaps . The newfound information will be helpful for public health and research efforts to understand, prevent , and manage OC. With a high number of annual live births (~240,000) and a diverse population, Florida provides a unique opportunity to study environmental determinant s of OC. T he main aims of this dissertation were : 1. To systematically review and perform a meta analysis on population based studies evaluating the associations between criteria air pollutants and OC to determine if there is sufficient evi dence to suggest a potential association . 2. To determine whether there was spatial clustering of OC in Florida from 200 4 20 08 , and investigate neighborhood level risk factors for OC clustering . i) Hypothesis 2.1 : T here was spatial clustering of OC in Florida during the study per iod after adjusting for significant known individual risk factors . ii) Hypothesis 2 .2 . Neighborhoods with high average annual air pollution concentrations and low SES were significantly associated with being inside an OC cluster. 3. To use a population based case cont rol study to examine the association between OC and exposures to PM 2.5 or O 3 during the first trimester of pregnancy , and to identify susceptible subpopulations among singleton live births in Florida from 2004 2008 . i) Hypothesis 3.1 : There was a positive association between exposure s to PM 2.5 or O 3 during the first trimester and OC .

PAGE 32

32 ii) Hypothesis 3.2 : Individual risk factors such as i nfant sex , maternal race, and maternal smoking during pregnancy modif ied these associations. Stated differently, th e purposes of this dissertation were to a ddress the following research questions: 1. What are the overall effect estimates for the association between criteria pollutants and OC based on existing literature? 2. Were there any clusters of high OC rates in the sta te of Flo rida from 2004 2008 , after adjusting for individual risk factors? 3. Which neighborhood level characteristics are associated with OC clustering? 4. Were there any associations between PM 2.5 or O 3 exposures during the first trimester and OC among singleton live births in Florida from 2004 2008 ; and, if so, what was the magnitude of these association s ? 5. Were these association s modified by individual risk factors including infant sex , maternal race, and maternal smoking status during pregnancy ? This dissertation is the first to utilize large population based data to evaluate the association between air pollution and OC while simultaneously identifying susceptible population s in the state of Florida . Additionally , it is one of the first to study potential spatial clustering and explore neighborhood level determinants for OC in the US and Florida . It offers several strengths. First, it provide s a cost effective way to study the association between air pollution and OC by using existing data. Second , the population based data will provide high representation of births in Florida. In addition, the identification of potential cluster s and susceptible populations is critically important for scientific and public health purposes, as it may provide import ant information for the further understanding of the underlying mechanisms , generation of hypothesis fo r environmental influences, and target sub groups for prevention of adverse health effects of air pollution.

PAGE 33

33 Furthermore, to overcome the limitations in the exposure assessment in previous studies, many of which relied on the closest monitor, this dissertation utilizes the EPA Hierarchical Bayesian modeling system (HBM) output. This model provides a more flexible and better resolution of exposure estimates because it incorporates air pollution monitors value s , the national emission inventory data, meteorological data , and photochemical properties of pollutants to predict the level of air pollut ion at any given place and time .

PAGE 34

34 Figure 1 1 . Theoretical framework guiding present research . Solid border boxes indicate domains of risk factors that are relevant for this research.

PAGE 35

35 Figure 1 2. Trend in studies on the association between air pollution and birth defects , 2000 2014

PAGE 36

36 CHAPTER 2 ASSOCIATION S BETWEEN AIR POLLUTION AND ORAL CLEFT S: A SYSTEMATIC REVIEW AND META ANALYSIS Introduction Birth defects are the leading cause of infant mortality in the U S and in many developed countries, accounting for more than 20% of all infant deaths ( Martin et al., 2008 ) . Oral clefts (OC) are among the most common types of major birth defects with a prevalence of 1.7 per 1 , 000 births ( Parker et al., 2010 ) , which is equivalent to approximately 7,100 cases annually in the US. O ral clefts are usually classified into two categories: cleft lip with or without cleft palate (CLP) and cleft palate only (CPO). Alt hough isolated OC are often not as fatal as other types of birth defect, they are associated with two times higher risk of infant mortality within the first year of life compared to the unaffected population (Carlson et al., 2013). In addition, they are as sociated with high morbidity . Specifically, i nfants affected by OC face significantly higher risk of severe recurring ear infections as well as feeding, hearing, speech, and dental complications ( Conrad et al., 2014 ; de Vries et al., 2014 ; Sharma and Nanda, 2009 ; Tannure et al., 2012 ) . Most cases of OC require multiple surgical interventions to minimize complications later in life. Thus, infants with OC require significantly higher cost of healthcare and management c ompared to those that are unaffected ( Boulet et al., 2009 ; Cassell et al., 2008 ; Wehby and Cassell, 2010 ; Wehby et al., 2012 ; Weiss et al., 2009 ; Yazdy et al., 2008 ) . O ral clefts ar e also associated with the high risk of long term disability ( Wehby et al., 2014b ) and lower quality of life for both the infants and their families ( Antunes et al., 2014 ) . Therefore, the understanding of etiologies for ef fective prevention of OC as well as other birth defects remains a priority in maternal

PAGE 37

37 and child health management and research, especially as the etiologies of most birth defects remain unknown. There is growing evidence suggesting a potential causal asso ciation between exposure to ambient air pollution and the risk of adverse birth outcomes including infant mortality ( Loomis et al., 1999 ; Ritz et al., 2006 ; Woodruff et al., 1997 ) , low birth weight ( Bobak, 2000 ; Romao et al., 2013 ) , preterm delivery ( Bobak, 2000 ) , and intrauterine growth restriction ( Pereira et al., 2012 ) . The a mbient air pollutants that have commonly been studied include particulate matter (PM) with aerodynamic diameter of less than 2.5 or 10 µm ( PM 2.5 or PM 10 ), carbon monoxide (CO), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), and ozone (O 3 ) . These pollutants are also called criteria air pollutants . They have been known to have s erious health effects , and are currently used by the US Environmental Protection Agency (EPA) to set criteria for emission and regulation for air quality standards . More recently, there has been an increase in the number of studies investigating the association between criteria air pollu tants and the risk of birth defects ( Vrijheid et al., 2011 ) . However, due to the d ifferences in the study design and methods of exposure and outcome assessment, it is difficult to assess the overall association between OC and specific air pollutants across the studies. Due to the significance of OC and the quickly evolving field of envi ronmental epidemiology, specifically air pollution studies, it is important to determine an overall association between criteria air pollutants and OC . Therefore, it is now timely to systematically examine and perform a meta analysis on population based st udies evaluating the relationship between air pollution and OC . This new evaluation, focusing on OC , will help to determine if the

PAGE 38

38 association between air pollution and OC is consistent across studies , and to identify specific knowledge gaps to guide new r esearch efforts. Materials and Methods Search S trategies A literature search of the fol lowing databases was conducted: PubMed, Pro Quest, Web of Science, Toxline, that me t the following criteria were identified: a) published in English, b) original epidemiologic study, c) defined OC as at least one of the stu dy outcomes, d) studied human prenatal exposure of specific criteria air pollutants including PM 2.5 , PM 10 , CO, SO 2 , NO 2 , or O 3 . We excluded studies that investigated the associations of acute/transient exposure as we are only interested in chronic ambient air exposures . Studies pertaining t o tobacco smoking or indoor pollution were not included . Studies that used ecologic exposure assignment (e.g. polluted area vs. unpolluted area) were not included due to high potential of misclassification bias. We also p erformed reference list searches for additional studies meeting our inclusion and exclusion criteria , and included studies published through May 2015. Meta analysis All studies meeting inclusion and exclusion criteria underwent a systematic qualitative review process that was completed by two independent reviewers. Meta analyses were performed to obtain a summary estimate of the association of each crite ria air pollutant with CLP or CPO. Analyses were conducted only if there were at least three studies r eporting a similar exposure and outcome combination. If a study

PAGE 39

39 included analyses with two overlapping populations, only the population with more participants was chosen for the analysis to avoid including the same sample more than once. Two types of summa ry risk estimates were calculated : one compared the risk at high vs. low level exposure and the other compared the risk for a unit increase in continuous exposure. For each of these calculations, we used risk estimates (odds ratio or relative risk) and their respective 95% confidence intervals (CIs) from the fully adjusted single pollutant model reported from each study. For the high vs. low exposure analysis, we selected risk estimates for the highest (e.g. fourth quartile) compared to the lowest category (e.g. first quartile) of exposure. If a study only provided risk estimates for continuous exposure, the estimates were conver ted to those for an interquartile range (IQR) increase in exposure for each study to facilitate comparability for the high vs. low level comparison . This also gives a more conservative estimation because the difference between IQR is smaller than that fro m the highest to the lowest quartile. For continuous exposure analyses, results were first converted into common units of exposure since studies reported different units for the same pollutant. For example, Schembari et al. (2013) reported their results fo r NO 2 3 ( Schembari et al., 2014 ) while other studies reported in parts per million (ppm) ( Marshall et al., 2010 ) , parts per billion (ppb) ( Padula et al., 2013a ) or parts per hundred million (pphm) ( Gilboa et al., 2005 ) . Results were converted into common u nit s using the following formulas : p pm = ( g/m 3 value )(24.45)/(MW) 3 = (ppm value)(MW)/24.45

PAGE 40

40 where 24.45 is a conversion factor that represents the volume of one mole of gas, and MW represents the molecular weight of the air pollutant at temperature of 25°C (77°F) and pressure of 1 atmosphere (760 torr or 760 mm Hg). Next, individual study effect estimates we re converted to that for a common increment in exposure. For SO 2 , some studies reported risk estimates for a 1 ppb increase in exposure ( Hwang and Jaakkola, 2008 ) ; however, another reported for a 0.6 ppb increase ( Hansen et al., 2009 ) . The final units of exposure increment that were used for analyses were as followed: 1 ppb for SO 2 ; 10 ppb for O 3 and NO 2 ; 10µg/m 3 for PM 2.5 and PM 10 ; an d 1 ppm for CO . Summary risk estimates were obtained using fixed effect or random effects model based on methodological variation in study approach and heterogeneity guided by the I squared (I 2 ) statistic ( Higgins et al., 2003 ) . The I 2 statistic represents the proportion of variation in the risk estimates that is due to heterogeneity rather than chance. It is calculated as: I 2 = [(Q df)/Q] 100% where Q is the chi squared statistic defined as Q = W i x (lnOR mh lnOR i ) 2 and df=k 1 represents its deg rees of freedom for k studies. Here, W i represents the inverse variance weight for each study, OR mh represents the Mantel Haenszel summary effect for all studies, and OR i represents the odds ratio for individual studies. When methodological approach was similar and I 2 value was low (<25%), indi cating low heterogeneity, fixed effects models were used to obtain summary estimates. Otherwise, random effect models were used to obtain the summary estimate s . The summa ry estimates were determined by weighted linear combination of individual studies

PAGE 41

41 weig hted by the inverse of within study variance (fixed effect analysis) , or both the between study and within study variances (random effect analysis). In addition, two sensitivity analyses were also performed. The first sensitivity analysis (S1) included onl y studies that report associations for first trimester exposure, and the second one (S2) included studies that only examined exposure during weeks 3 8. Gestational weeks 3 8 are known as the etiologically critical window for the potential effects of air po llution on birth defects ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Marshall et al., 2010 ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ) . Meta analysi s results were displayed using Forest plots. Lastly, publication bias analysis was performed using Egger test and funnel plots ( Egger et al., 1997 ) . If the resulting p value was less than 0.10 (indicating potential bias), the trim and fill method was used to obtain the adjusted odds ratio ( Duval and Tweedie, 2000 ) . All analyses were performed using Comprehensive Meta Analysis software version 2 (Biostat Inc., Englewood, NJ). Results Qualitative R esults Figure 2 1 illustrates the selection of relevant publications for this review . Over 8 , 4 00 articles were obtained from the original title search. After applying a priori inclusion and exclusion criteria, 36 abstracts were selected for full text review. After further excluding studies with ecological design (n=1), no OC as an outcome (n=13), or n o specific criteria air pollutant (n=10 ), the final review and analyses included 12 studies. Of these, six were conducted in the United States, two in the United Kingdom, one in Spain , one in Taiwan, one i n Israel, and one in Australia. Eight studies utili zed a case control study design with cases identified from routine birth defect registries and controls from birth registries or community hospitals. Of the remaining, two had a cohort

PAGE 42

42 design, one cross sectional design, and one with both cohort and case c ontrol design. Table 2 1 provides a detailed review of the specific characteristics of the studies included in this review. The studies differed with respect to the method of exposure assessment, outcome assessment, participant selection, and adjustment fo r potential confounding. A few studies directly relied on measurements at the closest available monitor for their exposures ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Marshall et al., 2010 ; Ritz et al., 2002 ) while o thers included averages of all monitors within certain distance ( Rankin et al., 2009 ) . Some used geostatistical techniques such as inverse distance w eighting ( Hwang and Jaakkola, 2008 ; Padula et al., 2013a ) , land use regression ( Schembari et al., 2014 ) or other models ( Dolk et al., 2010 ; Farhi et al., 2014 ; Vinikoor Imler et al., 2013 ; Vinikoor Imler et al., 2015 ) to estimate exposures. Most studies used average daily concentration as the exposure unit, except one that u sed average monthly concentrations ( Hwang and Jaakkola, 2008 ) . The levels of CO appeared t o be highest in southern California ( Ritz et al., 2002 ) with an IQR between 1.14 2.39 ppm compared to other places where the IQRs were mostly less than 1 ppm. For O 3 , the concentrations appeared to be high California, North Carolina , and Texas, USA ( Padula et al., 2013a ; Vinikoor Imler et al., 2013 ; Vinikoor Imler et al., 2015 ) compared to other locations. Daily NO 2 and particulate matter concentrations were gen erally higher in Taiwan, Spain and Israel compared to other countries. No study has reported air pollution concentration that was higher than

PAGE 43

43 Studies also varied with respect to the methods of participant selection in terms of the distance from the monitors ranging from 12 km 40 km ( Hansen et al., 2009 ; Marshall et al., 2010 ) ; others had no criteria in terms of this distance. Furthermore, a large variation in inclusion and exclusi on criteria was also observed. For instance, some studies only included live births ( Hwang and Jaakkola, 2008 ; Marshall et al., 2010 ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ) while others also considered fetal deaths after 20 weeks of gestation ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Ritz et al., 2002 ) , or terminations after 20 weeks of gestation or diagnosis of birth defect ( Dolk et al., 2010 ; Pad ula et al., 2013a ; Rankin et al., 2009 ) . Many studies used record linkages for their case and control definitions without confirmation of outcomes or any mention of quality control. Exceptions to this were the studies by Hwang et al. 2008 ( Hwang and Jaakkola, 2008 ) , Rankin et al. 2009 ( Rankin et al., 2009 ) and Padula et al. 2013 ( Padula et al., 2013a ) , all of which confirmed cases by clinical or autopsy reports. In terms of comparability of groups, studies either matched or adjusted for important confounders; however, variation exists on the types of confounders that were controlled for. Quantitative R esults Table 2 2 presents summary risk estimates of the association of different combination s of criteria air pollutants with OC . Meta analysis results indicated that when all eligible studies were included, there was no evidence of an association between any pollutant and CLP in both the categorical and continuous exposure analyses. However, there appeared to be an inverse association between CO and CPO (odds ratio [ OR ] : 0.77 ; 95% CI: 0.63 0.95) when the highest quartiles were compared to the lowest quarti les across studies (Table 2 for this association .

PAGE 44

44 Sensitivity analyses were also performed to explore consistency (Table 2 3). The S1 analysis only included studies that examined the association betwe en pollutants measured during first trimester, and the S2 analysis only included studies that reported on exposures during gestation weeks 3 8. In S1, the inverse association between CO and CPO remained consistent with cases having 23% decreased odds of be ing in the highest quartile of exposure compared to controls (OR: 0.77, 95% CI: 0.63 0.95) (Table 2 3, Figure 2 2) . However, in S2, SO 2 during weeks 3 8 of gestation was found to be associated with a 23% increased risk of CLP (OR: 1.23, 95% CI: 1.03 1.47) when comparing the highest quartiles to the lowest quartiles across studies (Table 2 3 and Figure 2 and funnel plots indicated no significant publication bias for these association s (Figure 2 4) . Discussion The purpose of this study was to evaluate existing evidence linking air pollution and OC . The meta analysis found that SO 2 exposure from weeks 3 8 of gestation was significantly associated with an increased risk of CLP. On the other hand, CO exposure during the first trimester was found to be inversely associated with CPO. No other significant association was found for any other pollutant outcome combination. However, results should be interpreted with caution sin ce data were based on a small number of existing studies with significant methodological issues that are worthy of discussion. Participant Selection As previously mentioned, s tudies differed greatly on methods of participant selection. Differences in distance from the air monitors may affect the results due to local variation in air pollution concentration . The longer the distance from the air

PAGE 45

45 monitors , the higher likelihood that the effects of air pollution are less pronounced due to poorer accuracy . However, in their sensitivity analyses examining the effects of air pollution at different concentric circles, Hansen et al. 2009 reported that the effects of air pollution are reduced when people living close r to air monitors were included km vs. km) ( Hansen et al., 2009 ) . This phenomenon needs further investigat ion in future studies. Case selection often come s from routine birth defec t registries and controls from Vital S tatistics. As quality control and validation studies for birth defect registries and birth registries are still lacking, this method still has so me considerable weaknesses due to potential misclassification resulting from underreporting or delayed diagnosis, especially in developing countries. Case definition was also different across studies. For instance, some studies only included live births ( Hwang and Jaakkola, 2008 ; Marshall et al., 2010 ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ) while others also considered still births or fetal deaths after 20 weeks of gestation ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Ritz et al., 2002 ) , or terminations after 20 weeks of gestation or diagnosis of birth defect ( Dolk et al., 2010 ; Padula et al., 2013a ; Rankin et al., 2009 ) . Some severe cases , who may be high ly exposed, may not become live births . T herefore, the inclusion of only live births may systematically exclude exposed cases , leading to potential underestimation of association. Exposure A ssessment Most studies relied on ex isting networks of ambient air monitors to quantify exposure. Some studies quantif ied exposures at maternal address at delivery based on measurements at nearest monitor ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Marshall et al., 2010 ; Ritz et al., 2002 ) , while others characterized the exposure based on averages

PAGE 46

46 of all monitors within a certain distance ( Rankin et al., 2009 ) . Desp ite good feasibility, this method suffers from serious limitations. Relying on the few available monitors is likely to result in exposure misclassification because monitors are relatively sparse and are likely to be insufficient for capturing all the spati al variations in air pollution. In addition, this method does not capture the residential mobility or daily activities patterns of the mother during pregnancy. In order to address issues discussed above, interpolation or geostatistical methods have been d eveloped to more accurately estimate air pollution exposure. These methods use complex mathematical models to predict pollution levels at unmeasured locations using measured samples at various locations. A method that was commonly used in OC studies is inv erse distance weighting ( Hwang and Jaakkola, 2008 ; Padula et al., 2013a ) . This method gives more estimation flexibility, and enables the researcher to adjust for some spatial variation. Howev er, the model is still limited due to strict assumptions such as isotropy (e.g. the spatial variation is constant). In addition, inverse distance weighting may not be accurate when there are clustering effects. Some other methods that were used include lan d use regression model ( Schembari et al., 2014 ) and , more recently, outputs from o ther prediction models such as hierarchical Bayesian p rediction models ( Dolk et al., 2010 ; Vinikoor Imler et al., 2013 ) . The hierarchical Bayesian prediction model may give better spatial prediction as it incorporates parameters from many sources including air monitor records, the national emission inventory , meteorology, and chemi cal properties. Nevertheless, all of these methods still rely on indirect measurements of non personal data.

PAGE 47

47 Comparability and Confounding Confounding in observational research can threaten study validity in many ways, particularly because the associations of air pollution with birth outcomes are likely s mall compared to other factors such as demographics, nutrition , and genetics. Most studies o n birth defects rely on information on birth certificates or birth defect registr ies . Birth records usually have s ufficient demographic information; however, have limited medical/clinical or lifestyle information. The studies we reviewed used different data sources; therefore, we were unable to control for different groups of confounders, making it difficult to comprehensively compare results. Some studies attempted to residence ( Gilboa et al., 2005 ; Hansen et al., 2009 ) . This method may help minimize some potential confounders related to geographic location; however, this may bias th e results towards the null because it reduces the exposure differences between cases and controls. In addition, residual confounding, which may occur due to imperfect classification of confounders, needs particular consideration. A crucial element in air pollution research is personal activity patterns. The studies we reviewed geocoded maternal residence at birth or fetal death based on record. Therefore, none were able to control for residential mobility. Even if the mother stay ed at the same address thro ugh pregnancy and birth, her daily activity patterns may have expose d her to pollution concentrations different from those predicted by the stationary air monitors or mathematical models. Therefore, future studies may need to control for daily activity pat terns in order to minimize potential misclassification of exposure.

PAGE 48

48 Biologic Plausibility A few mechanisms were suggested to support biological plausibility of the association between early exposure to SO 2 and the risk of CLP found in this study. It has be en shown that air pollutants can invade the body directly through diffusion or active/passive transport, and can induce responses including inflammation, oxidative stress, changes in cardiovascular functions, and endocrine disruption ( Brook and Rajagopalan, 2012 ; Peters et al., 1997 ) . These changes have been hypothesized to affect the maternal fetal exchange mechanism and increase the risk of adverse birth outcomes ( Slama et al., 2008 ) . However, their effects on the risk of OC are still speculative and un clear. The observed inverse association of CO with CPO may initially seem counterintuitive given the well known severe adverse effects from high CO concentrations resulting from its high affinity to hemoglobin (Hb) ( Dubrey et al., 2015 ) . However, studies have shown that in a healthy non smoking person, the concentration of carboxyhemoglobin ( CO Hb ) is around 0.5 1.5% ( EPA, 2010 ) . Among healthy individuals, e xposure to low concentrations of CO in ambient air induces CO Hb concentration only within the normal margin of Co Hb. In fact, research has consistently shown that exposures to CO at lower concent rations may even have beneficial health effects ( Cai et al., 2015 ; Durante et al., 2006 ; Ryter et al., 2006 ) . For example, low concentrations of CO have been shown to have anti inflammatory, anti oxidative, and anti microbial effects ( Chin and Otterbein, 2009 ; Klimova et al., 2013 ; Otterbein et al., 2000 ; Otterbein et al., 1999 ; Otterbein et al., 2005 ; Pae et al., 2004 ; Wilson et al., 2012 ) . More importantly, there is also evidence showing that low CO can have similar effects in the intrauterine environment, and thus may also have protective effects during

PAGE 49

49 pregnancy ( Klimova et al., 2013 ; Olgun et al., 2014 ) . It i s also important to no t e that the observed inverse association between CO and CPO was held significant only by a single study ( Marshall et al., 2010 ) . In this study, the authors discussed that misclassification could have been a limitation because participants were up to 40 km away from air moni tors ; in a sensitivity analysis, the findings were st ill consistent when it was restricted to participants within 10 km from the monitors . More research is still needed to investigate this observation. Limitations The results of this meta analysis should be interpreted with cautions due to several limitation s. De spite the advantage of a large pooled sample size, the results , specifically significant findings, were still based on a small number of studies that differed with respect to study design, selection of participants, and assessment of exposure and outc ome as discussed above . Additionally, although significant associations were found for the categorical exposure analyses comparing highest versus lowest quartiles, no associations were detected in the continuous exposure analyses. It is unclear which method of exposure modeling is more appropriate given their individual limitations . For the continuous exposure analyses, it is assumed that there is a log linear relationship between the pollutants and OC , which may not actually be accurate . For the categ orical exposure analyses, results were pooled over study areas with very different exposure level s . Furthermore, the composition of air pollution may vary across different study areas. These differences may lead to different relationships between pollutant s and OC , and may be affected by the less polluted study areas resulting in diluted overall estimates . Lastly, although we have tested for pot a low power test. Therefore, w ith the few

PAGE 50

50 studies included in our ana lyses, we may not have enough power to detect publication bias. Future D irection s Based on our discussion, a few methodological issues would need to be addressed in the future studies. These include , but are not limited to the following recommendations. Th e use of personal air monitors providing real time individual level exposure data is a preferred method of exposure assessment that reduces misclassification. This would allow researchers to consider different aspects of the exposure including concentratio n, time, and dose ; this would lead to better cumulative and time specific (e.g. gestational window) exposure measurements. Despite advantages of this direct exposure assessment method, it may not be feasible given the cost and rarity of OC. Therefore, the development of more spatially and temporally sensitive model for better exposure assessment would be more cost efficient. There is also a strong need to investigate the combined effects of different pollutants. This may be an important direction because ai r pollution is composed o f many different constituents. Air pollution research on other adverse birth outcome has identified that certain subgroups based on infant sex and maternal race may be more susceptible ( Bell et al., 2008 ; Medina Ramon and Schwartz, 2008 ) . Ho wever, no studies have attempted to identify susceptible subpopulations for the association between air pollution and OC . The answers to this question a re important for public health interventions as well as the understanding of the biologic mechanism ; the refore, it is important for future studies to identify potential susceptible subpopulations. Similarly, w hile air pollution is likely to have positive associations with birth defects, genetics is likely to have a gre ater

PAGE 51

51 influence. Thus, t he gene environment interaction is vital and should be investigated in future studies . Furthermore, it is also important to design studies to understand the underlying mechanisms that are responsible for the effects of air pollution and birth defects which could h elp to develop more effective preventive strategies . Meanwhile, according to the EPA, PM 2.5 and O 3 are two of the most detrimental air pollutants for health . However, as demonstrated in this chapter, there is currently no consistent evidence on the associ ation between these pollutants and OC. This underscores the need for more stud ies focusing on these two air pollutants as there are currently only five existing studies ( Marshall et al., 2010 ; P adula et al., 2013a ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ; Vinik oor Imler et al., 2015 ) . The next two chapters of this dissertation will focus on these two criteria air p ollutants. Conclusion This meta analysis of 12 studies investigating the associations of criteria air pollution with OC found that while SO 2 exposures during early pregnancy may increase the risk of OC, this may not be the case for CO. No associations were detected between any other pollutant OC combinations. Results need to be interpreted with caution as the meta ana lysis includes relatively few st udies with some degree of heterogeneity. More studies are needed to better understand the association between air pollution and OC and the underlying biologic mechanisms.

PAGE 52

52 Table 2 1. Characteristics of studies included in the review References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Periods of Exposure Confounders C onsidered Ritz et al. 2002 California, USA, 1987 1993 Case control 3121 cases: cardiac or oral facial cleft that were live births or fetal death >20 weeks of gestation; 9357 controls: randomly selected from all births and fetal deaths Mean 24 hr measurements at nearest station within 16km p25 p75: CO: 1.14 2.39 ppm O 3 : 1.07 2.86 pphm Continuous Months 1,2,3 Decade born, infant sex, maternal race, age, single vs. multiple birth, parity, prenatal care, maternal educati on, season of conception and other pollutants Gilboa et al. 2005 Seven counties in Texas, USA, 1997 2000 Case control 1719 cases: cardiac or oral clefts who were live births or fetal deaths >20 weeks of gestation 3667 controls: live birth or fetal death randomly selected from all live births and fetal deaths; frequency matched by vital status, year, and maternal county of residence at delivery Mean hourly or daily concentration at the nearest or next nearest station from residence. Average nearest distanc e: 8.6 14.2 km (max: 36 54 km) p25 p75: CO: 0.4 0.7 ppm NO 2 : 1.3 2.1 pphm O 3 : 1.8 3.1 pphm PM 10 :19.5 29 3 SO 2 : 1.3 2.7 ppb Categorical Weeks 3 8 of pregnancy Alcohol consumption, attendant of delivery, gravidity, material status, maternal age, education, illness, race/ethnicity; parity, place or delivery, plurality, prenatal care, season of conception, and tobacco use.

PAGE 53

53 Table 2 1. Continued References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Perio ds of Exposure Confounders C onsidered Hwang et al. 2008 Taiwan 2001 2004 Case control 653 cases: live births with cleft lip with or without palate 6530 controls: randomly selected from all live births Monthly average concentration measured by inverse distance weighting from 72 monitoring stations in Taiwan p25 p75: CO: 0.48 0.76 ppm NO x : 16.0 23.9 ppb O 3 : 24.4 30.1 ppb PM 10 : 44.8 64.5 3 SO 2 : 2. 4 5 .0 ppb Continuous Months 1,2,3 Maternal age, plurality, gestational age, population density, season of conception Rankin et al. 2009 Northern UK 1985 1990 Case control 2779 cases: had congenital anomalies, gestational weeks when terminated, miscarriage live, or still births 15000 controls: randomly selected from all live and stillbir ths Mean daily concentrations from all monitor stations within 10 km p25 p75: SO 2 : 2.7 3 Categorical Trimester 1 Birth weight, sex, maternal deprivation

PAGE 54

54 Table 2 1. Continued References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Periods of Exposure Confounders C onsidered Hansen et al. 2009 Brisbane, Australia, 1997 2004 Case control 245 cases: with oral cleft defects, still and live births; matched 1:5 with controls age, marital status, indigenous status, number of previous pregnancies, month of last menstrual period , area level SES, distance to pollution monitor. Daily a verage concentration of PM 10 , NO 2 , SO 2 , and 8 hr average of CO and O 3 at closest monitor site Mean: CO: 1.1, ppm NO 2 : 8.2, ppb O 3 : 25.8 , ppb PM 10 : 18.0 3 SO 2 : 1.5 ppb Continuous Weeks 3 8 Match criteria: maternal age, marital status, indigenous status, parity, month of last menstrual period, area level socioeconomic status, distance to monitor; and neonate sex Dolk et al. 2010 England, 1991 1999 Cross section al 586 cases of cleft lip with or without cleft palate and 302 cases of cleft palate amo ng 759,993 live or still births, or fetal deaths termination after diagnosis of defect Annual mean concentration measured by model output with resolution 1x1 km grids p10 p90: NO 2 3 PM 10 3 SO 2 3 Continuous Average of year 1996 Maternal age, deprivation index, region, hospital catchment

PAGE 55

55 Table 2 1. Continued References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Periods of Exposure Confounders C onsidered Marshall et al. 2010 New Jersey, USA, 1998 2003 Case control 717 cases: cleft lip with or without cleft palate or cleft palate, live births; 12925 controls: normal births randomly selected from all births Average hourly or daily averages measured at the nearest station. Average distance: 13 20 km; max: 40 km p25 p75: CO: 0.65 1.02 ppm NO 2 : 0.02 0.03ppm O 3 : 0.0 2 0.03 ppm PM 2.5 : 20 .0 30 .0 3 PM 10 : 22 .0 33.5 3 SO 2 : 0.003 0.007 ppm Categorical Weeks 3 8 of pregnancy Race, age, education, gravidity, alcohol use, smoking, season of conception, infant sex Padula et al. 2013 California, USA, 1997 2006 Case control 806 cases: live, still births, terminated pregnancy with defects; 849 controls: non malformed, live born randomly selected from birth hospit als. Average daily 24 hr concentration of NO 2 , NO, CO, PM 10 , PM 2.5 and 8 hr max concentration of O 3 measured by inverse distance weighting from up to four nearby stations. p25 p75: CO: 0.40 0.72ppm NO: 4.15 20.2 ppb NO 2 : 13.4 20.5ppb O 3 : 29.1 62.7 ppb PM 2.5 : 10.9 26.1 3 PM 10 : 25. 3 44. 1 3 Categorical Months 1, 2 Maternal race/ethnicity, education, vitamin use Vinikoor Imler et al. 2013 North Carolina, USA, 2003 2005 Cohort 395 cases with oral cleft defects among a cohort of 322,969 live births Avera ge 24 hr concentrations predicted by the EPA CMAQ Bayesian models at 12x12km resolution. Mean ( IQR): O 3 : 40.7(19.1 ) ppb PM 2.5 : 14.0 (3.9 ) 3 Continuous Weeks 3 8 of pregnancy Race, maternal age, rural urban continuum codes category, co pollutant effects

PAGE 56

56 Table 2 1. Continued References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Periods of Exposure Confounders C onsidered Schembari et al. 2014 Barcelona, Spain,1994 2006 Case control 2247 cases of non chromosomal defect s who are live, still weeks or terminated after diagnosis of defect 2991 controls free of defects (continuously selected by random date of birth as 2%) who were live births Average concentrations predicted by land use r egression models. Mean (IQR): NO 2 3 NO x 3 PM 10 3 PM coarse 3 PM 2.5 3 PM 2.5 absorbance: 2.65 (0.73) 10 5 /m Continuous Weeks 3 8 of pregnancy Maternal age, conception season, year of birth/termination, SES index. Farhi et al. 2014 Israel, 1997 2004 Cohort 4058 infants with congenital anomalies among 216,730 infants (207,825 natural conception infants and 8905 ART infants) Kriging using monthly average for 117 air monitor stations across the country Mean (SD): NO x : 26.3 (16.6) ppb O 3 : 32.1(4.5) ppb PM 10 3 SO 2 : 2.74(0.75) ppb Continuous and categorical Trimester 1,2,3 Maternal ethnicity, maternal country of birth, maternal education, mode of conception, plurality, season of birth, infant sex .

PAGE 57

57 Table 2 1. Continued References Settings Study Design Participants Exposure A ssessment Exposure Range and T ype Critic al Periods of Exposure Confounders C onsidered Vinikoor Imler et al. 2015 Texas, USA, 2002 2006 Mixed method (cohort and case control) 21,060 cases of birth defects among 1,422,671 births in Texas, and 291 cases of birth defects among 812 controls free of birth defects Average 24 hr concentrations predicted by the EPA Bayesian models at 12x12 km resolution. Mean (SD): O 3 : 40.0 (8.8)ppb PM 2.5 3 Continuous Trimester 1 P renatal care during the first trimester, number of previous live births, maternal age, maternal race/ethnicity, and maternal education, urbanity, folic acid/multivitamin use, smoking, and alcohol consumption during the month prior to conception through the first trimester . Abbreviation: ART, assisted reproductive technology; CMAQ, community multi scale air quality; EPA, Environment Protection Agency; IQR, interquartile range; LMP, last menstrual period; SC, spontaneously conceived; SES, socioeconomic status; SD, standard deviation.

PAGE 58

58 Table 2 2 . Summary risk estimates of the as sociations between criteria air pollutants and oral c lefts *indicates statistical significance at 0.05. a Studies included: 1) Ritz et al. 2002, 2) Gilboa et al. 2005, 3) Hwang et al. 2008, 4) Rankin et al. 2009, 5) Hansen et al. 2008, 6) Dolk et al. 2010, 7) Marshall et al. 2010, 8) Padula et al. 2013, 9)Vinikoor Imler et al. 2013, 10) Schembari et al. 2014, 11) Farhi et al. 2014, 12) Vinikoor Imler et al. 2015. Studies 4, 7, and 8 (indicated as bold) were not included in continuous analyses. b Th e summary ORs are for specific increment in the concentrations of pollutants: 1 ppm CO, 10 pp b NO 2 , 10 ppb O 3 , 10 3 PM 2.5 and PM 10 , 1 ppb SO 2 c If I 2 was <25 .0, results from fixed effect models were presented; otherwise, results from random effects models were presented. d value was < 0.10 then the summary OR was adjusted using trim and fill method; otherwise, an unadjusted summary OR was used. Pollutants Studies included a Number cases Categorical (high vs. low) Continuous b Summary OR (95% CI) c, d Heterogeneity (I 2 ) Publication bias p value Summary OR (95% CI) c, d Heterogeneity (I 2 ) Publication bias p value Cleft lip with or without cleft palate (CLP) CO 1,2,3,5, 7, 8 2 , 151 0.94(0.75,1.17) 76.2 0.33 0.81(0.54,1.19) 66.9 0.61 NO 2 2,5,6, 7,8 ,10 1 , 799 1.00(0.85,1.19) 42.5 0.15 1.01(0.99,1.16) 0.00 0.46 O 3 1,2,3,5, 7,8 ,9,11,1 2 4 , 612 1.03(0.97,1.11) 10.4 0.18 1.00(0.95, 1.06) 8.94 0.25 PM 2.5 7,8 ,9,10,12 2 , 846 0.97(0.90,1.05) 22.3 0.05 1.04(0.97, 1.11) 0.00 0.37 PM 10 2,3,5,6, 7,8 ,10,11 2 , 531 1.00(0.93,1.07) 0.00 0.16 1.00(0.95,1.04) 0.00 0.97 SO 2 2,3, 4 ,5,6, 7 ,11 2 , 337 1.06(0.89,1.27) 64.0 0.25 0.98(0.95,1.01) 26.4 0.28 Cleft palate only (CPO) CO 1,2,5, 7,8 795 0.77(0.63,0.95)* 37.3 0.41 0.91(0.70,1.18) 0.00 0.74 NO 2 2,5,6, 7,8 919 0.81(0.65,1.01) 11.0 0.01 0.97(0.67,1.41) 47.3 0.46 O 3 1,2,5, 7,8 ,9 983 1.00(0.95,1.05) 0.00 0.46 1.00(0.97,1.03) 0.00 0.48 PM 2.5 7,8 ,9 505 0.93(0.78,1.10) 0.00 0.13 Not enough studies PM 10 2,5,6, 7,8 929 0.89(0.70,1.13) 25.0 0.23 0.72(0.41,1.26) 61.9 0.31 SO 2 2,5,6 ,7 764 0.92(0.74,1.15) 41.2 0.39 1.02(0.97,1.07) 0.00 0.44

PAGE 59

59 Table 2 3. Sensitivity analyses results for summary risk estimates of the associations between criteria a ir pollutants and oral c left s *indicates statistical significance at 0.05 a Sensitivity analyses: S1, Restricted to only first trimester exposure; S2, Restricted to only weeks 3 8 exposure. b Studies included: 1) Ritz et al. 2002, 2) Gilboa et al. 2005, 3) Hwang et al. 2008, 4) Rankin et al. 2009, 5) Hansen et al. 2008, 6) Dolk et al. 2010, 7) Marshall et al. 2010, 8) Padula et al. 2013, 9)Vinikoor Imler et al. 2013, 10) Schembari et al. 2014, 11) Farhi et al. 2014, 12) Vinikoor Imler et al. 2015. Studies 7 and 8 (indicated by bold) were not included in continuous analyses. Pollutant outcome combination Sensitivity analysis a Studies included b Number cases Categorical exposure Continuous exposure c Summary OR (95% CI) d ,e Summary OR (95% CI) d, e Cleft lip with or without cleft palate (CLP) CO S1 1,2,3,5, 7,8 2 , 151 0.94(0.75,1.17) 0.81(0.54,1.19) S2 2,5, 7 845 1.01(0.59,1.75) Not enough studies NO 2 S1 2,5, 7,8 ,10 1 , 213 1.01(0.83,1.24) 1.03(0.88,1.20) S2 2,5, 7 ,10 967 1.10(0.96,1.23) 1.03(0.88,1.20) O 3 S1 1,2,3,5, 7,8 ,9,11,12 6 , 412 1.04(0.97,1.11) 1.00(0.95, 1.06) S2 2,5, 7 ,9 1 , 078 0.90(0.78,1.04) 0.97(0.89,1.07) PM 2.5 S1 7,8 ,9,10,12 2 , 846 0.95(0.87,1.05) 1.04(0.97, 1.11) S2 7 ,9,10 605 1.01(0.85,1.19) Not enough studies PM 10 S1 2,3,5, 7,8 ,10,11 1 , 945 1.00(0.93,1.08) 1.00(0.95,1.05) S2 2,5, 7 ,10 863 1.04(0.93,1.15) 1.08(0.93,1.25) SO 2 S1 2,3,4,5, 7 ,11 1 , 751 1.12(0.92,1.36) 0.99(0.92,1.07) S2 2,5, 7 737 1.23(1.03,1.47)* Not enough studies Cleft palate only (CPO) CO S1 1,2,5, 7,8 795 0.77(0.63,0.95)* 0.91(0.70,1.18) S2 2,5 ,7 508 0.71(0.47,1.05) Not enough studies NO 2 S1 2,5, 7,8 617 0.81(0.60,1.11) Not enough studies S2 2,5, 7 507 0.80(0.53,1.21) Not enough studies O 3 S1 1,2,5, 7,8 ,9 983 1.00(0.95,1.05) 1.00(0.97,1.03) S2 2,5, 7 ,9 667 0.95(0.78,1.17) Not enough studies PM 2.5 S1 7,8 ,9 505 0.93(0.78,1.10) Not enough studies S2 7 ,9 400 Not enough studies Not enough studies PM 10 S1 2,5, 7,8 627 0.92(0.69,1.23) Not enough studies S2 2,5, 7 501 0.94(0.63,1.39) Not enough studies SO 2 S1 2,5, 7 462 0.82(0.66,1.01) Not enough studies S2 2,5, 7 462 0.82(0.66,1.01) Not enough studies

PAGE 60

60 c T he summary ORs are for specific increment in the concentrations of pollutants: 1 ppm CO, 10 ppb NO 2 , 10 ppb O 3 , 10 3 PM 10 and PM 2.5 , 1 ppb SO 2 . d If I 2 w as we re <25 , results from fixed effect models were presented; otherwise, results from random effects models were presented. e value was <0.10 then the summary ORs were adjusted using trim and fill method; otherwise, unadjusted summary ORs were presented.

PAGE 61

61 Figure 2 1. Selection of relevant studies included in the study

PAGE 62

62 Figure 2 2. Forest plot of summary estimate of the association between CO exposure during first t rimester and CPO. Individual study point estim ates were for comparison between the highest vs. the lowest quartiles.

PAGE 63

63 Figure 2 3. Forest p lot of s ummary e stimate of the a ssociation b etween SO 2 e xposure d uring w eeks 3 8 and CLP. Individual study point estimates are for comparison between the highest vs. the lowest quartiles.

PAGE 64

64 Figure 2 4 . Funnel plots for the associations between CO exposure during first trimester and CPO (A), and SO 2 exposure during weeks 3 8 of gestation and CLP (B).

PAGE 65

65 CHAPTER 3 LOCALIZED CLUSTERING AND NEIGHBORHOOD LEVEL DETERMINANTS OF ORAL CLEFT S IN FLORIDA Introduction Oral clefts (OC ) are among the most commo n type s of birth defects in the US as well as around the world . They have a combined prevalence of 1.7 per 1000 birth s , which is equivalent to approximat ely 7 , 100 cases annually in the US ( Parker et al., 2010 ) . Although OC are often not as fatal compared to some ot her types of birth defects, they are associated with significant morbidity and financial burden. Children with OC face significantly higher risk for complications related to feeding, speech, hearing, recurrent ear infections, as well as dental development ( Conrad et al., 2014 ; de Vries et al., 2014 ; Sharma and Nanda, 2009 ; Tannure et al., 2012 ) . Many affected children need multiple surgical interventions to prevention developmental complications. Consequently , infants affected by OC re quire significantly higher expenditure for healthcare and management compared to those who are unaffected . The mean cost of hospitalizations associated with OC from birth to tw o years of age for each case is approximately $21,090 while the mean cost for ho spitalizations among infants w ithout craniofacial defects is about $2,504 ( Weiss et al., 2009 ) . In addition, affected infants may also have a higher risk of long term disabilities compared to tho se unaffected ( Moraleda Cibrian et al., 2014 ; Mossey and Modell, 2012 ; Wehby et al., 2014a ) . Despite evidence of significant decline in the rates of birth defect s after the fortificatio n of folic acid in the national food supplies in the late 1990s , this decline is mostly for neural tube defects ( CDC, 1992 ; CDC, 1993 ; CDC, 2004 ; Honein et al., 2001 ; Williams et al., 2005 ) . Consequently, the rest of birth defect cases including OC are likely influenced by other factors, most of which are unclear . More importantly , there is

PAGE 66

66 still evidence of significant disparities in the prevalence of OC with respect to demographic and geogra phic characteristics ( Canfield et al., 20 06a ; Poletta et al., 2007 ) . Consequently, further understanding of the et iology for in order to direct efforts towards prevention of OC remain important priorities in maternal and child health management and research . The etiology of OC remains unclear; however, it is hypothesized to be caused by a combi nation of individual and environmental factor s as discussed in Chapter 1. Oral clefts have been associated with maternal demographic factors such as infant sex and maternal race ( Lebby et al., 2010 ) . They have also been linked to a variety of behavio ral risk factors during pregnancy including maternal smoking ( Chung et al., 2000 ; Lie et al., 2008 ; Zhang et al., 2011 ) , alcohol consumption ( Boyles et al., 2010 ; Grewal et al., 2008 ) , maternal nutrition ( Figueiredo et al., 2015 ; McKinney et al., 2013 ) , as well exposures to some drugs ( Lin et al., 2012 ; Mines et al., 2014 ) . More recently, exposure to var ious air pollutants including particulate matter with aerodynamic diameter les s than 2.5 or 10 µ m ( PM 2.5 or PM 10 ) and ground level ozone (O 3 ) have also been associated with OC ( Farhi et al., 2014 ; Hwang and Jaakkola, 2008 ; Marshall et al., 2010 ; Vinikoor Imler et al., 2013 ; Vrijheid et al., 2011 ) . However, these associations are still inconsistent in the literature as illustrated in Chapter 2. The risk factors for OC may be unevenly distributed across ge ographic area s . Thus, geographical variation of OC rates may also exist. Therefore, investigation of the spatial patterns of OC using techniques that involve spatial cluster detection can be of great public health value for identifying areas with e levated risk, and for further understanding of the etiology of OC . A cluster is defined as an area that has a number

PAGE 67

67 of cases that is significantly higher than that expected if cases are randomly distributed throughout the whole study region . Identification of spa tial clusters , especially after adjustment for individual level risk factors, may reveal potential environmental etiologies and allows generation of hypothesis about underlying environmental determinants . Furthermore, id entification of high risk cluster is important to aid with health services planning, allocation of resources, and other pr evention and management efforts . O ral clefts are particularly appropriate to this type of study because the latency period between environmental exposures and the develop ment of the defect is relatively short . This minimizes the potential bias associated with residential mobility during exposure period and allows for stronger hypotheses about the environmental factors that may be associated with it . For this reason, a few recent studies have investigated potential clustering of birth defects including gastroschisis, a serious abdominal wall defect ( Root et al., 2009 ; Yazdy et al., 2015 ) . In addition, some studies have found evidence of OC clustering in South America ( Poletta et al., 2007 ) . However, to our knowledge, no studies have investigated potential clustering of OC in America. Besides from cluster detection, studying the characteristics of th e neighborhoods that are part of an OC cluster is also important in order to explore the reasons behind the observed high risk . The identification of neighborhood level determinants using neighborhoods as unit s of analysis can also help in identifying high risk communities for provision of needs based health services . For these reasons, studies have identified the impact of neighborhood characteristics on many health outcomes ( Yen et al., 2009 ) including cardiovascu lar disease ( Chum and O' Campo, 2013 ; Sarrafzadegan et al., 2012 ) , cancer ( Gomez et al., 2012 ; Shariff Marco et al., 2014 ) , mental health ( Truong

PAGE 68

68 and Ma, 2006 ) , and adverse birth outcomes ( English et al., 2003 ) . However, those for birth defects and specifically OC received much less attention in the literature . Florida is a state with relatively large number of live birth s annually (n~240,000) , an d a diverse population . This makes it a unique setting for studying potential clustering of OC and the associated neighborhood risk factors . Thus, t he first purpose of this study was to determine whether there was spatial cluster ing of OC in the state of F lorida among live births born to residents from 2004 2008 . After OC clusters are identified, w e further aim ed to determine neighborhood risk factors for OC by compar ing the social and chemical environment indicators between census tract block groups that fell within the cluster (s) and those that did not . As previously mentioned, exposures to PM 2.5 and O 3 have been associated with birth defects in the literature ( Vrijheid et al., 2011 ) . However, as demonstrated in Chapter 2, this association is still inconsistent for OC . Therefore, we aim to determine whether neighborhood level mean annual co ncentrations of PM 2.5 and O 3 are associated with OC clustering . Secondly, we are also interested in neighborhood socioeconomic status (SES) indicators that have been commonly studied for adverse birth outcomes ( Blumenshine et al., 2010 ) . The findings of this study may potentially be important for etiologic research of OC , as well as resource/service allocation and policy decisions at the local level that are crucial for addressing neighborhood health disparities. Materials and Methods Setting and Participants All live births born from 2004 20 08 to Florida residents were identified from Florida Vital Statistics ( N= 1,125,374 ) . We excluded multiple gestations because we did not want to include the same pregnancy twice, and that they shared the intrauterine

PAGE 69

69 environment and genes (n= 35,394 ). Of the rema ining births , 37,098 births had missing information on whether they had an OC and were not included in the analysis. The births were geocoded at residential address at delivery by using ArcGIS (ESRI, Redlands, CA) or R package ggmap , which is based on the Google Map Application Programming Interface ( Kahl e and Wickham, 2015 ) . Of these, we were unable to geocode 113 births due to invalid addresses . The final analysis included 1,052 , 769 live births . The outcome, OC, was determined using electronic birth certificates data from Florida Vital Statistics. OC cases (n= 636 ) and non cases were aggregated to census block group (CBG) level defined by the 20 00 US Census . A CBG is the smallest unit with information provided by the Census and is compr ised of between 600 3000 people . Cluster A nalysis In order to determine spatial clustering of OC , several covariates that may have the potential to explain spatial clustering of OC were selected. These covariates were selected based on several reasons: a) they appear to have an associa tion with OC in the st udy population , and b) they may cluster in space and can potentially explain OC clustering if it exist s . These covariates include infant sex (female, male), gestational complications ( y es, n h igh school, > h igh school ), maternal coun try of origin (US, non US), maternal race (White, Black, Hispanic, Other) , and maternal sm oking during pregnancy ( y es, n o) . These covariates were also assessed from birth certificates. spatial scan statistics were used to detect whether there was any spatial clustering of OC during the study period ( Kulldorff, 1997 ) . At each CBG , ular window increasing in size that moves in space (Figure 3 1 ). In effect, it obtain s an infinite number of overlapping circles of

PAGE 70

70 different sizes , jointly covering the entire study region, where each circle reflects a possible cluster . The radii of the circles vary continuously in size from zero to a user defined maximum, which was origin ally set to 50% of the total Florida live birth population during the study period . In subsequent analyses, i t was apparent that the maximum circle size of 20% of the population produced the same findin gs; it was then set to 20% to reduce computational tim e. In the se analyses, it is assumed that the number of OC cases in each CBG follow s a Poisson distribution. At each window, the test statistic tests for the nu ll hypothesis that the risk of OC is the same inside compared to outside of the circle. The procedure finds out whether the number of observed cases inside a circle exceeds the expected number. Here the expected number of cases is calculated as: E[c]= p*C/P where c is the number of cases in the circle, p is number of live births in the circle, C is the total number of cases in the population, and P is the total live births in Florida. For each window , the like lihood of finding the observed number of OC and the expected number under the null hypothesis is calculated . The likelihood function is maxi mized over all window locations and sizes; t he circle with the maximum likelihood is the most likely cluster, that is, the cluster least likely to be due to chance . The likelihood ratio for this window constitutes the maximum likelihood ratio test statisti c. Test statistic p values were achieved by Monte Carlo simulation of the data under the null hypothesis with 999 permutations.

PAGE 71

71 The scan procedure was performed unadjusted and adjusted for the aforementioned covariates. In the adjusted analyses, for each covariate category i, the expected number of cases was calculated as: E[c] = E[c i ] = p i * C i / P i where c is the number of cases in the circle , C is the total number of cases in Florida , and P is the total live births in Florida. Each covariate was introduced into the model one at a time to assess for their individual effect s . From a computational standpoint, we were unable to adjust for more than two variables simultaneously due to small number of cases in each CBG. This does not allow us to partition our cases into different covariate categories as the p value is unstable when there are too many cells with no information ( Kulldorff, 1997 ) . Cluster analysis was performed using SatScan version 9. 3 ( SatScan, Cambridge, MA ) . Neighborhood Predictors of Oral Cleft Clustering We also aim ed to determine neighborhood level determinants of clustering by comparing the characteristics of CBGs that fell within the clusters to those who were outside of the clusters. Several potential neighborhood level risk factors were determined. A ir pollution data were obtained from the US Environmental Protection Agency (EPA) National Environmental Public Health Tracking networ Bayesian Prediction Model (HBM) output. The methodology for the HBM model is described elsewhere ( McMillan, 2010 ) . The HBM output includes 12x12 km gridded estimates of PM 2.5 (daily average) and O 3 (daily 8 hr maximum) surfaces. The PM 2.5 is measured in mass units of micrograms per cubic meter (µg/m 3 ) and O 3 is measured in parts per billion (ppb). For the purpose of this study, we extracted data f or the state of Florida during the period 2003 2008. The year 2003 was included because this year

PAGE 72

72 overlaps with pregnancy time for some births in 2004. In order to determine air pollution level for each CBG , a CBG layer was overlaid with the 12x12 km gridd ed estimates for the two air pollutants. A CBG is assigned the annual average concentration of PM 2.5 or O 3 of the 12x12 grid in which it falls . Each pollutant measure was analyzed as a continuous variable. Additional variables were also obtained from the 20 00 US Census. We specifically included six neighborhood socioeconomic status ( SES ) indicators that have been commonly investigated in studies of adverse birth outcomes ( Blumenshine et al., 2010 ) . These SES variables include CBG level percent of White population, percent of resident s age 25 or older with bachelor degree or more , percent of population living below federal poverty level , median family income , percent of population age 16 and above who are employed, and percent rental occupancy . These variables represent different SES domains that associate with health: education, poverty, racial composition, housing, and employment. Furthermore, we also assessed whether a CBG fell in an urban or rural area. S ocioeconomic status measure s were an alyzed as continuous variables. In addition, while some researche rs have advocated the use of composite SES index to study the cumulative effects of different SES measures ( Messer et al., 2008 ) , others have shown that single separate measures do not differ ( Krieger et al., 2003b ) . Therefore, we chose to employ both methods in this study. To construct a composite SES index, we followed the approach from several previous studies ( Carmichael et al., 2009 ; Carmichael et al., 2003 ; Krieger et al., 2003a ; Wasserman et al., 1998 ) . Socioeconomic status indicators were divided into quartiles based on the distribut ion among the CBGs . For each neighborhood measure, a score of

PAGE 73

73 1 was assigned to the high risk quartile and 0 otherwise. Specifically, for percent employment, percent of residents with degree or more , percent of White population, and median family income, the lowest quartile was assigned as a 1 and 0 otherwise . However, for percent living under f ederal poverty level, and percent rental occupancy, the top quartile was assigned a value of 1, and 0 otherwise. The neighborhood SES score for a CBG was then calculated as the sum of these values across the six SES measures. The cut offs at the 25th and 7 5th percentiles for the SES indicators were as followed: $ 39,904 and $ 75,250 for median family income , 5.3% and 19.8% for percent poverty, 8.2% and 26.1 % for p or more , 4 6.6% and 66 . 7 % for percent 16 and older employed, 69.1% a nd 95.2% for percent White, and 60% and 92.9% for percent rental housing , respectively . C ensus block group s were divided into two groups : those that fell within the detected clusters (cluster=1) and those that did not (cluster=0). We defined cluster based on the model that was adjusted maternal race and maternal smoking since these variables explained the most number of excess cases in the unadjusted model. We first used descriptive statistics to describe the distribution of charac teristics between the groups. After that, u nivariate and multivariate logistic regression model s were used to determine the associat ions of CBG level air pollution and SES indicators with OC clustering. Results Cluster A nalysis Originally CLP, CPO, and both OC outcomes combined were analyzed separately . No significant spatial clustering of CPO was observed , and clustering of CLP was similar to that of both types combined . Therefore, we proceeded with the

PAGE 74

74 analysis for both OC outcomes combined . Table 3 1 present s the results of the scan statistics. In the unadjusted analysis, three spatial clusters were identified (Figure 3 2 A) . Cluster 1 was located in the northwest Panhandle region of Florida covering parts of Okaloosa and Walton County . This cluster had 23 cases and a risk of 2.58 times higher than the rest of the state. Cluster 2 was located in the north central part of the state, which covered Putnam, Marion, Lake, Volusia, and Sumter C ounty. This cluster had a total of 112 cases with an expected number of 43.2 cases, yielding a relative risk of 2.93 compared to the rest of the state . Cluster 3 was located in the southwest central region covering Hillsborough, Polk, Manatee, Hardee, De S oto, Highlands, and Sarasota County . This cluster contained 1 22 cases, which yielded a rate of approximately 1. 7 5 times higher than the rest of the state. After adjustment for infant sex , gestational complications, or maternal education, the results remain ed the same, suggesting that the spatial distribution of these variables did not explain the excess cases in the observed clusters (Table 3 1 and Figure s 3 2 B D ). When adjusted for maternal country of origin, C luster 3 became a little smaller in size , but the estimates remained consistent ( Figure 3 3 A). When maternal race was added to the m odel, C luster 2 no longer exist ed , and C luster 3 became much smaller (Figure 3 3 B ), suggesting that the spatial distribution of maternal race further explained some exces s in the number of OC cases. When race and maternal smoking were in the model , C luster 1 did not change; however, the second cluster became smaller, suggesting that maternal smoking also explained some of the excess in the number of cases in this region (Figure 3 3 C ) . Neighborhood Predictors of Oral Cleft Clustering Table 3 2 describes the univariate differences in characteristics between CBGs that fell inside a cluster and those that did not. When comparing cluster CBGs to

PAGE 75

75 non cluster CBGs, significan t differences were observed with respect to air pollution levels as well as other characteristics including median family income, education level, and rurality. Specifically, compared to CBGs outside of the clusters, those that fell within cluster had sign ificantly higher annual levels of PM 2.5 (11.2 vs. 9.7 µg/m 3 , p<0.001) and O 3 (40.2 vs. 36.9 ppm, p<0.001). C ensus block groups inside the cluster also had lower percent of adults with a degree or more, and lower median family income (Table 3 2). Percent White, percent renter occupied housing , percent under federal poverty level, and percent employment were not associated with clustering. When using composite SES score, those with a score of 3 or more (indicat ing lower SES ) appeared to be more likel y to fall inside a cluster ; however , this was not statistically significant. Lastly, rural CBGs were significantly more likely to fall into a cluster compared to urban CBGs. Table 3 3 presents the unadjusted and adjusted associations between CBG char acteristics and high risk OC clustering. In the unadjusted analyses, CBG annual PM 2.5 and O 3 concentrations were significantly associated with OC clustering. Specifically, 1 µg/m 3 increase in annual PM 2.5 concentration was associated with a 52% increased r isk of being i n an OC cluster. Similarly, one ppb increase in annual O 3 concentration was asso ciated with a 63% increased risk of being in an OC cluster. Percent adult age 25 or older with degree , percent White population, and median family income were marginally and negatively associated with OC clustering. When composite SES score was used, those with a score of 3 or more had a 27 % increased risk of be ing in a cluster compared to those with a score of 0. However, this

PAGE 76

76 association was not si gnificant. In addition, rural CBG had more than 11 times increased risk of being inside a cluster. A fter adjustment for each other, separate SES indicators, and rurality, a unit increase in annual PM 2.5 and O 3 concentration was associated with a 41% and 2 8% increased risk of being inside a cluster, respectively. Rural CBGs remained associated with more than 10 times the risk of being inside a cluster compared to urban areas after adjustment for air pollution concentration and SES indicators . The associatio ns between the individual SES indicators as well as their composite score and OC cluster became insignificant a fter adjustment for urbanity status and air pollution concentrations ( Table 3 3) . Median family income was marginally significant with 7% decreas ed risk for a $10,000 increase . Discussion Our results provide evidence supporting that spatial clustering of OC may have existed in Florida during the study period. We also found that average annual concentrations of PM 2.5 and O 3 concentration, and rurality were positively associated with OC clustering. These findings fill an important gap in the literature. A lthough some studies have shown spatial clustering of other birth defects including gastroschisis, OC received little atten tion despite being the most common form of major birth defect s ( Root et al ., 2009 ) . Our study is first to simultaneously explore spatial clustering and neighborhood risk factors for OC clustering. The identification of the two poten tial OC clusters with significantly elevated risk compared to expectation is of great public health interest, especially for Florida. Although the clusters did not contain the majority of the cases in Florida, they suggest that OC occur at much higher rate s than expected even after adjustment for some significant individual risk factors in these two regions. This

PAGE 77

77 may suggest some environmental contributors that need further investigation some of which we have also explored in this study. For example, our ec ologic analyses suggest ed that high PM 2.5 and O 3 exposure and rurality may partially contribute to higher risk of OC. These findings are consistent with other studies, which found evidence that exposures to PM 2.5 and O 3 were associated with OC at the indiv idual level ( Farhi et al., 2014 ; Hwang and Jaakkola, 2008 ; Marshall et al., 2010 ) . However, since this is an ecologic association, individual level study is necessary to confirm these associations in F lorida (see Chapter 4). Our results showing that rurality was highly associated with OC clustering is consistent with the literature for OC and other birth defects ( Langlois et al., 2010 ; Langlois et al., 2009b ; Li et al., 2013 ; Messer et al., 2010 ) . The specific reasons underlying the association between rurality and OC is not clear. However, it is possible that women from rur al areas may be exposed to a very different environment compared to those in the urban areas; thus, are affected differently. For example, since Florida has high agriculture land coverage, women in the rural areas may be more exposed to agriculture related risk factors that may increase their risk of having infants affected by OC ( Engel et al., 2000 ; Heeren et al., 2003 ; Langlois et al., 2009b ; Ochoa Acuna and Carbajo, 2009 ) . In addition, women in rural areas may have li mited access to certain health related amenities and resources that may affect their risk ( H artley et al., 1994 ; Slifkin et al., 2004 ) . However, this is a speculation and needs further investigation. Our results showi ng lack of evidence of association between neighborhood SES indicators and OC are consistent with current literature. Specifically, while some studies found that neighborhood SES deprivation was associated with the risk of some other

PAGE 78

78 birth defects, such as sociation for OC was not apparent ( Carmichael et al., 2003 ; Vrijheid et al., 2000 ) . This lack of association may be explained purely by the spatial nature of this study. This means that the true SES indicators for the CB Gs may have changed over time during the study period. However, we were not able to adjust for temporal change, leading to a lack of association. More studies are needed to further investigate these findings. Meanwhile, besides from air pollution and rural ity, it is important to investigate whether there are other contributors for elevated risk of OC in the clusters. This study is limited by several shortcomings. T he OC clusters detected in this study are approximations of the true clusters. This means tha t while we detected the general location of the cluster, we are uncertain about the exact boundaries of these clusters . W e cannot infer that the women living within the cluster area s are at the same risk. As discussed in Chapter 1, women have varying levels of risk, which depend on their individual characteristics. However, the presence of the cluster s after adjustment for significant individual risk factors suggests that additional risk factor such as environment al may exist in these area s . W e cannot exclude the possibility that other risk factors may also partially be responsible for the observed clustering. The association between air pollution and OC cluster in this study is purely sp atial, and we only used average annual concentrations to serve as a proxy for actual exposures. However, the levels of air pollution may vary with time and may affect the risk of OC differently based on time of exposure for each pregnancy. This study was u nable to capture temporal variation in air pollution due to the purely spatial nature. We address ed this issue in Chapter 4 of this dissertation.

PAGE 79

79 Another limitation involves the fact th at this geographic analysis use d residence at birth. This means that w e assumed that women stayed at the same residence throughout pregnancy. However, studies have shown that between 25 and 30 percent of women change d residence between conception and birth ( Fell et al., 2004 ; Khoury et al., 1988 ) . A majority of the relocations appear ed to be local ( Fell et al., 2004 ; Khoury et al., 1988 ) and the characteristics of women who move we re similar to those who d id not ( Canfield et al., 2006b ) . S ince we used CBG as a unit of analysis, residential mobility may not have significantly affected our study. In addition, although the scan method used in this study adjusts for multiple testing using data simulation, the clusters we detected may have been spur ious. Lastly, birth certificates are not an ideal source to capture birth defects cases as they are not as accurate as birth defect registries. However, this is more applicable in birth defects that are invisible at birth and are often diagnosed after birt h (e.g. congenital heart defects). Since most OC are visible forms of birth defect and are detected prenatally, it has been shown that birth certificate has good completeness with respect to OC may provide useful information ( Wang et al., 2006 ) . Nevertheless, it is important to keep in mind that some cases may have been misclassified as non c ases. If this w ere true, then our results would have been biased depending on where the misclassified cases were located. Conclusion We found potential clustering of OC in Florida from 2004 2008 after adjusting for significant individual covariates. This suggests potential additional environmental determinants, some of which we have also demonstrated in the study. Specifically, we found a consistent positive association between neighborhood concentrations of PM 2.5

PAGE 80

80 and O 3 as well as rurality with OC after a djusting for important individual risk factors. Thus, we cannot dismiss these findings as a chance occurrence. More studies are still needed to confirm the association and explore the mechanism of the association. Our next chapter will further investigate the associations between PM 2.5 as well as O 3 and OC on the individual level.

PAGE 81

81 Table 3 1. Spatial cluster analysis of oral clefts in Florida, 2004 20 08 Abbreviations: RR, relative risk within the cluster compa red to the rest of Florida ; LLR, log likelihood ratio ; CBG, census block group. Covariates Cluster #CBG Radius (km) Births Cases Expected RR LLR p value Unadjusted 1 168 76.1 15,078 23 9.1 2.58 14.7 0.004 2 667 63.0 71,547 112 43.2 2.93 13.0 0.013 3 1,316 94.4 122 , 425 119 74.1 1.7 5 13.3 0.011 Infant sex 1 168 76.1 15,078 23 9.1 2.58 14.6 0.004 2 667 63.0 71,547 112 43.2 2.93 12.9 0.013 3 1,316 94.4 122 , 425 119 74.1 1.7 5 13.3 0.011 Gestational complications 1 168 76.1 15,078 23 9.1 2.58 14.6 0.004 2 667 63.0 71,547 112 43.2 2.93 12.9 0.013 3 1,316 94.4 122 , 425 119 74.1 1.7 5 13.3 0.011 Maternal education 1 168 76.1 15,078 23 9.1 2.58 14.6 0.004 2 667 63.0 71,547 112 43.2 2.93 12.9 0.013 3 1,316 94.4 122 , 425 119 74.9 1.7 2 13.3 0.010 Maternal country of origin 1 168 76.1 15,078 23 9.1 2.58 14.6 0.004 2 667 63.0 71,547 112 43.2 2.93 12.9 0.013 3 1,184 91.7 102,686 104 62.4 1.80 13.1 0.008 Maternal origin, race 1 169 76.1 15,090 23 9.1 2.58 13.1 0.016 2 61 21.3 5,935 13 3.5 3.68 11.7 0.040 Race, maternal smoking 1 168 76.1 15,078 23 9.1 2.58 13.0 0.016 2 41 14.9 4,885 10 2.9 3.43 11.0 0.030

PAGE 82

82 Table 3 2. Neighborhood characteristics of census block groups inside and outside of oral cleft clusters in Florida, 2004 20 08 Characteristics Inside cluster Outside cluster p value a N % Mean SD N % Mean SD Total 209 2.3 8,916 9 7.7 Average annual PM 2.5 (µg/m 3 ) 11.2 1.0 9.7 1.6 <.0001 Average annual O 3 (ppb) 40 .0 1.2 36.9 3.4 <.0001 Census block group family income 56,964.3 22 , 592.0 61,014.2 32,858.1 0.0117 Percent below federal poverty level 14.9 9.9 14.3 12.7 0.4911 Percent White 74.6 27.5 77.5 24.4 0.0977 16.6 14.0 18.6 13.6 0.0344 Percent employed 54.6 17.2 55.5 15.8 0.4224 Percent renter occupied housing 72.7 24.2 71.8 27.7 0.5833 Composite SES score 0 45 21.5 1 , 935 21.7 0.1461 1 56 26.8 2 , 960 33.2 2 57 27.3 2 , 301 25.8 51 24.4 1 , 730 19.4 Urbanity Rural 177 84.7 2 , 880 32.3 <.0001 Urban 32 15.3 6 , 036 67.7 a p values are for t test if variable is continuous, and chi square test if variable is categorical. Abbreviation: N, number; PM 2.5 , particulate matter with aerodynamic diameter less than 2.5 micrometer; O 3 , ozone; S D, standard deviation ; SES, socioeconomic status .

PAGE 83

83 Table 3 3. Associations between neighborhood characteristics and oral cleft cluster s in Florida, 2004 20 08 Characteristics Odds ratio (95% CI) Unadjusted Adjusted a Average annual PM 2.5 ( 1 µg/m 3 ) 1.52 (1.43, 1.62) 1.41 (1.31, 1.51) Average annual O 3 ( 1 ppb) 1.63 (1.51, 1.77) 1.38 (1.27, 1.50) Census block group family income ( $10,000) 0.96 (0.92, 1.00) 0.93 (0.85, 1.00) Percent below federal poverty level (5%) 1.0 2 (0.9 7 , 1. 07 ) 1.05 (0.98, 1.12 ) Percent W hite (5%) 0.98 (0.95, 1.00) 0.97 (0.94, 1.00) Percent with (5%) 0. 94 (0.89, 1.00 ) 1.0 2 (0.9 5 , 1. 08 ) Percent employed (5%) 0. 98 (0.94, 1.03 ) 0.98 (0.91, 1.08) Percent renter occupied housing (5%) 1.01 ( 0.98 , 1. 0 3 ) 1.00 (0.97, 1.02) Neighborhood SES score 0 1.00 1.00 1 0. 81 (0. 55 , 1. 21 ) 0. 81 (0. 54 , 1. 23 ) 2 1. 07 (0. 72 , 1. 58 ) 1. 11 (0. 73 , 1. 68 ) 1. 27 ( 0.85 , 1.90 ) 1. 38 ( 0.9 0 , 2. 10 ) Urbanity Urban 1.00 1.00 Rural 11.6 (7.9,16.9) 10. 8 (7. 3 , 16.0 ) Abbreviation: CI, confidence interval; PM 2.5 , particulate matter with aerodynamic diameter less than 2.5 micrometer; O 3 , ozone; SES, socioeconomic status . a Adjusted for the other pollutant, the six SES indicators, and urbanity

PAGE 84

84 Figure 3 1. Spatial scanning method using . CBG stands for census block group.

PAGE 85

85 Figure 3 2. Spatial c luster s of oral cleft s in Florida, 2004 20 08 . The panels represent clusters unadjusted for any covariate (A) , adjusted for infant sex (B) , gestational complications (C) , and maternal education (D) .

PAGE 86

86 Figure 3 3. Spatial clusters of oral clefts in Florida, 2004 2008. Panels represent cluster adjusted for maternal country of origin (A), maternal country of origin and maternal race (B), and maternal race and maternal smoking (C).

PAGE 87

87 CHAPTER 4 ASSOCIATIONS BETWEEN PRENATAL EXPOSURES TO AIR POLLUTION AND ORAL CLEFT S Background Oral cleft s ( OC ) are the most common types of non chromosomal birth defects with a prevalence of 1.7 per 1000 births ( Parker et al., 2010 ) , which is equivalent to about 7,100 cases annually in the United States ( US ) . Although OC are often not as fatal compared to some other types of birth defects including neural tube defects, children affected by OC face a significantly higher burde n of morbidity . Specifically, they have a higher risk of complications related to feeding, speech, ear infections, and dental development. Most require multiple surgical interventions to prevent subsequent complications. As a result, affected children need considerably higher cost of healthcare and management compared to those who are unaffected ( Boulet et al., 2009 ; Cassell et al., 2008 ; Wehby and Cassell, 2010 ; Wehby et al., 2012 ; Weiss et al., 2009 ; Yazdy et al., 2008 ) . Oral clefts are also associated with higher risk of long term disability ( Wehby et al., 2014b ) and lower quality of life for both the infants and the ir families ( Antunes et al., 2014 ) . Despite evidence of significant decline in the rates of birth defects after the fortification of folic acid in the national food supplies, this decline was observed only for neural tube defects ( Honein et al., 2001 ) . Consequently, the rest of birth defect cases including OC are likely influenced by other unknown factors. Currently, there is no known effective intervention for OC. On top of that, there is still evidence of significant disparities in the prevalence of OC with respect to demographics an d geography ( Canfield et al., 2006a ; Poletta et al., 2007 ) . Thus, further understanding of the etiolog y

PAGE 88

88 of OC rem ains an important priority in maternal and child health management and research . According to the US Environmental Protection Agency ( EPA ) , particulate matter (PM) and ground level ozone (O 3 ) are among air pollutants that have the most significant effects on human health. There is mounting evidence linking PM of various sizes and O 3 to many serious health endpoints. These endpoints include mortality ( Schwartz, 1994 ) , respiratory diseases ( Laumbach and Kipen, 2012 ) , cardiovascular diseases ( Gill et al., 2011 ) , cancer ( Raaschou Nielsen and Reynolds, 2006 ) and many others ( Cho et al., 2014 ) . Moreover, exposures to PM with aerodynamic diameter of 2.5 or 10 micrometer (PM 2.5 or PM 10 ) or O 3 have been asso ciated with the risk of adverse birth outcomes such as infant mortality ( Ritz et al., 2006 ) , low birth weight ( Bobak, 2000 ) , preterm delivery ( Bobak, 2000 ) , and intrauterine growth restriction ( Pereira et al., 2012 ) . Consequently, there have been concerns that exposure to PM and O 3 may also increase the risk of birth defects including OC . However, there are a limited number of studies investigating this association. Moreover, this small body of literature shows mixed evidence with some studies suggesting positive association ( Hwang and Jaakkola, 2008 ; Marshall et al., 201 0 ) , and some no association ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Padula et al., 2013b ; Rankin et al., 2009 ; Ritz et al., 2002 ) . This has also been demonstrated in Chapter 2 of this dissertation. The discrepancy in findings is likely partially attributable to misclassification of exposure. M any existing studies on the association between air pollution and OC directly rely on measurements at nearest station ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Marshall et al., 2010 ; Ritz et al., 2002 ) or the average of all monitors within a

PAGE 89

89 certain distance ( Rankin et al., 2009 ) for exposures. While convenient, this method suffers from serious limitations. Specif ically, relying on the few available monitoring stations is likely to result in exposure misclassification because the monitoring network monitors are relatively sparse and are insufficient for capturing all the spatial variations in air pollution. For exa mple, the real exposure at residence is likely different from the closest monitor, which could be tens of kilometers away. Furthermore, infant and maternal chara cteristics including infant sex, maternal smoking during pregnancy , and maternal race have bee n widely suggested as strong risk factors for OC ( Mossey et al., 2009 ) . Exposure to air pollution and smoking may have synergic effects on health ( Turner et al., 2014 ) . In addition, studies on air pollution and adverse birth outcomes (e.g. low birth weight and preterm delivery) have also suggested that these associations ma y differ among infants of different sexes ( Ghosh et al., 2007 ) , and that the health effects of air pollution may differ by race/ethnicity ( America n Lung Association, 2001 ; Sexton et al., 1993 ) . However, our literature review on the association between air pollution and OC revealed no existing study that examined susceptible subgroups based on these characteristics . Such information is important for the identification of target group for intervention , and for the further understanding of the biologic mechanism . Furthermore, in Chapter 3, we have demon strated an ecologic association between PM 2.5 and O 3 and OC risk in Flor ida from 2004 2008. Therefore , there is an immediate need to confirm this association at the individual level. This population based case c ontrol study aim ed to address some shortcomin gs from previous studies. First, in order to overcome the issues related to lack of spatial

PAGE 90

90 resolut ion, we examine d the association between exposure to PM 2.5 and O 3 during specific exposure windows of pregnancy and OC using the US EPA National Environmenta l Public Health Tracking networ (HBM) output ( McMillan, 2010 ) . This model predicts the concentrations of specific air pollutants at un known locations using information from different sources including air monitors, the N ational Emission I nventory, meteorological data as well as photochemical properties of particles to ensure improved spatial accuracy . Seco nd , we also aim ed to assess whet her the association of PM 2.5 and O 3 with OC differ ed by infant and maternal characteristics . Materials and Methods Setting and P articipants The source population include d all singleton live births born in Florida from January 1, 2004 to December 31, 2008. Births were identified from the Florida Vital Statistics data. Cases were defined as any live singleton birth who 1 ) was born to a mother who was a Florida resident at th e time of birth, 2 ) had an OC clearly defined on birth certificate without other known defects , 3 ) had a valid gestational age that is between 20 42 weeks. For each case , four controls were randomly selected among healthy births from the source population matched on date of birth. They had the same eligible criteria as cases, except they were healthy births with no birth defects . Since OC are rare outcome s , a control to case ratio of 4 :1 was chosen to optimize statistical power within an efficient design . E xposure A ssessment Air pollution data was obtained from the US EPA National Environmental Public Health Tracking networ The HBM combines PM 2.5 and O 3 data from the

PAGE 91

91 3/Community Multi scale Air Quality Model (CMAQ), which is based on the National Emission Inventory and meteorological as well as geographical factors. The methodology for the HBM model is described elsewhere ( McMillan, 2010 ) . The HBM model output includes 12x12 km gridded estimates of PM 2.5 (daily average) and O 3 (daily 8 hr maximum) surfaces. PM 2.5 is measur ed in mass units of micrograms per cubic meter (µg/m 3 ) and O 3 is measured in parts per b illion (ppb). For the purpose of this study, we extracted data for the state of Florida during the period 2003 2008. To obtain exposure for cases and controls, we geo residential address at delivery and overlaid this layer with the HBM output layer. Individual exposure during pregnancy was then estimated using daily concentrations in the grid in which the residential address falls. Since OC develop w ithin the first trimester of pregnancy, exposure was assessed during this time window. We determined specific exposure periods using gestational age given on birth certificate data. On Florida birth certificates, gestational age in weeks is determined by u ltrasound measurements. When ultrasound is not available, fundal height determined by clinical examination or menstrual history is used to estimate gestational age. Exposures were calculated as daily concentrations averaged over each a priori determined sp ecific window: weeks 1 2, 3 8, and 9 13 of gestation ; and the average concentration over the first trimester . Outcome A ssessment The main outcome s of this study are OC, which were categorized into two groups: cleft lip with or without palate (CLP) and cleft palate only (CPO). These outcomes were identified by Florida Vital Statistics birth data from 2004 2008.

PAGE 92

92 Covariates In order to account for potential confounders, additional variables were selected from birth certificates to include in the analyses . These variables included infant sex (female, m ale), maternal race (White, Black, Hispanic, Other), materna l age (<20, 20 34, 35), maternal education (high school or less, college educat ion), parental marital status (single/widowed, married), parity (0, , mate rnal smoking during pregnancy (yes, n o ), pregnancy alcohol use (yes, n o), pre pregnancy body mass index ( BMI ) (underweight, normal weight, overweight, o bese), gestational complications (yes, n o), census block group (CBG) level income (< state median ), urbanity (urban, rural) , maternal country of origin (US, non US), and gestational age in weeks (continuous ) . Gestational complications were defined as either gestational diabetes or gestational hypertension. Census block group incom e and urbanity were determined using the Census 2000 data. Statistical A nalysis We first conducted descriptive analyses and chi square test to compare characteri stics between cases and controls. C onditional logistic regression models were used to assess t he associations between air pollution and CLP as well as CPO. The main exposures (independent variables) were average daily PM 2.5 and O 3 over specific exposure windows as continuous variables. Odds ratios (OR) and 95% confidence interval s (CI s ) were obtain ed for a unit incre ase in each pollutant . We also assessed potential effect modificat ions of infant sex, maternal smoking during pregnancy , and maternal race on the associations between the air pollutants and OC . For each window of exposure, we introduced an interaction term between air pollution exposures and each potential effect modifier in the unadjusted model . For each comparison, an alpha

PAGE 93

93 level of 0.05 was used to assess statistical significance. All adjuste d models controlled for potential confounders described above. All g eographic information system ( GIS ) procedures were performed using ArcGIS version 10.2 (ESRI, Redlands, CA) and other statistical procedures were done in SAS version 9.4 ( SAS Institute, Ca ry, NC) . We performed two different sensitivity analyses to assess the robustness of our data. First, we fitted a multi pollutant model to determine the association of one pollutant with OC after adjusting for the other pollutant and covariates . Second, w e also linked our participants to the closest air monitor station. We then assumed that the partici that recorded at the closest air monitor and repeated the analyses. A map of PM 2.5 and O 3 monitoring station s during the study period are illustrated in Figure 4 1. Results A total of 2,524 controls were matched to 631 cases of OC , of which 478 were CLP and 153 were CPO. Table 4 1 presents characteristics of cases and controls and the associated chi square p values . Compared to the controls, the percent of male infants was higher among CLP cases (59.6% vs. 50. 1 % , p=0.0002 ); however, was lower among CPO cases (42.5% vs. 51.5 % , p=0.0 467 ). In addition, both CLP (59.2% vs. 45.7% , p <0.0001 ) and CPO (62.1% vs. 44.4%, p=0.00 12 ) cases were more likely to have mothers who were White compared to controls. Cleft lip with or without cleft palate cases also had a higher percentage of mothers who smoked during pregnancy (14.2% vs. 8. 6 % , p=0.0 002 ), lived in a rural area (12.3% vs. 8.2 % , p=0.00 42 ), or were born in the US (74.5% vs. 68. 5 % , p=0. 0112 ). Cleft palate only cases were more likely to have mothers who had gestational complications (13.1% vs. 7.0 % , p=0.0 283 ). Table 4 2 presents the distri bution of estimated air pollution exposures by exposure windows for

PAGE 94

94 cases and control s . In general, there were little differences in exposure between cases and controls. The Pearson correlation coefficients for PM 2.5 and O 3 over all exposure windows ranged from 0.31 for the first trimester average to 0.4 2 for weeks 1 2 . We detect ed significant interaction between PM 2.5 exposure and maternal smoking during pregnancy in the unadjusted model , but not infant sex or maternal ra ce . This interaction was significant for CLP during some exposure windows (e.g. weeks 3 8 of ges tation, p=0.0 2 ; first trimester, p=0.04 ). Therefore, we present both combined and stratified associations by maternal smoking during the pregnancy ( Table 4 3 ) . Exposure s to PM 2.5 during weeks 3 8 of gestation was associated with CLP (OR: 1.04, 95% CI: 1.01 1.07 for one 3 increase ) . However, the association was stronger among infants exposed to maternal smoking, with a 12% increased risk of CLP for 1 3 increase in exposure (OR: 1.12, 95% CI: 1.01 1.24). The association among those without maternal smoking during pregnancy was weaker and not significant (OR: 1.05; 95% CI: 1.00 1.12 for one 3 increase in exposure ). This association remained consistent after adjust ing for potential confounders (adjusted OR: 1.05, 95% CI: 1.00 1.09). O ne 3 increase in PM 2.5 exposure during weeks 3 8 of gestation was assoc iated with an 11% increased risk (adjusted OR: 1.11, 95% CI: 1.01 1.25) among infants exposed to m aternal smoking during pregnancy . Among those unexposed to maternal smoking during pregnancy, the association was weaker and not statistically significant (OR: 1.06, 95% CI: 1.00 1.13 for one 3 increase ) . No significant association was detected for any other exposure window for PM 2.5 and CLP. In addition, there was no evidence of association between PM 2.5 and CPO.

PAGE 95

95 For O 3 , there was significant interaction with in fant sex at alpha <0.05 during all exposure windows for CLP (p values for interaction were 0.03 for weeks 1 2 and 3 9, 0.01 for weeks 9 12, and 0.004 for the first trimester). In the unadjusted analyses, O 3 exposure was positively associated with CLP for all exposure windows for both sexes (Table 4 4) . H owever, the associations were more pronoun ced among females compared to males . After adjustment for potential confounders, the associations were still significant and stronger among females than males over all exposure windows . Among males, the associations for most windows became insignificant, except for during weeks 3 8 of gestation. For example, one ppb increase in O 3 exposure during weeks 3 8 of gestation was associated with a 7% (adjusted OR: 1.07, 95% CI: 1.0 3 1.10) incre ase in CLP risk among females, but only a 3% among males (adjusted OR: 1.0 3 , 95% CI: 1.01 1.0 6 ) (Table 4 4). The associatio n between O 3 and CPO was also significant over all exposure windows in the unadjusted analyses (Table 4 4) . There was no evidence of interaction with any effect modifier tested . In the adjusted analyses, the associations remained consistent during most exposure windows except weeks 9 13 . S ensitivity analyses involving multi pollutant models as well as using closest air monitoring stati on for exposure yielded mostly consistent results. When adjusted for O 3 , the association between PM 2.5 and CLP became statistically insignificant (Table 4 5). However, when adjusted for PM 2.5 , the association between O 3 and OC remained consistent (Table 4 6). When using closest air monitors as exposure assessment, the association between PM 2.5 and CLP remained consistent ( Table 4 7) . For O 3 , the positive associations with OC also remained mostly consistent (Table 4 8).

PAGE 96

96 D iscussion Our study found that PM 2.5 exposure during weeks 3 8 of gestation was significantly associated with increased risk of CLP. Th is association was stronger and significant among infants of mothers who smoked during pregnancy. Additionally, we found that O 3 exposure during the first trimester was also positively associated with both CLP and CPO. Females seemed to be more sensitive to the effects of O 3 on CLP . These results add to the limite d literature on the association between air pollution and OC . Relatively few studies have examined the associations of PM 2.5 or O 3 with OC in the literature ( Gilboa et al., 2005 ; Hansen et al., 2009 ; Hwang and Jaakkola, 2008 ; Marshall et al., 2010 ; Padula et al., 2013b ; Ritz et al., 2002 ; Vinikoor Imler et al., 2013 ) . Among these, no study has sp ecifically reported a significant association between PM 2.5 and OC ; h owever, many have reported that PM 2.5 exposure is associated with increased risk of many oth er types of major birth defect s ( Farhi et al., 2014 ; Padula et al., 2013b ; Schembari et al., 2014 ) . Maternal smoking has also been a well known risk factor for birth defects and OC ( Leite et al., 2014 ; Lie et al., 2008 ; Nieuwenhuijsen et al., 2013 ; Zhang et al., 2011 ) . Both exposures to smoking and air pollution may affect endocrine functions, increase oxidative stress and systemic inflammation among pregnant women ( CDC, 2010 ; Lee et al., 2011 ; Slama et al., 2 008 ) . These responses have been suggested to affect fetal maternal exchanges, which in turn may influence the risk of adverse birth outcomes and fetal development ( Kannan et al., 2006 ; Slama et al., 2008 ) . Therefore, it is biologically plausible that there may be synergi sti c effects between exposures to PM 2.5 maternal smoking during pregnancy . This speculation is consistent with our results which suggested that the association betwee n PM 2.5 and CLP was significant and stronger

PAGE 97

97 among those exposed to maternal smoking. However, given the results are new, more studies are need ed to confirm this finding . Fu rthermore, the fact that we found significant association during weeks 3 8 of gesta tion adds to the specificity of this association. This time window is known as the critical period for the lip and pala te during embryonic development , and has been used in many environmental exposure studies for OC as seen in Chapter 2, as well as other b irth defects ( Vrijheid et al., 2011 ) . The lack of association between PM 2.5 and CPO in this study is unclear. More studies are needed to confirm this observation as we had a l imited sample size of CPO cases. The positive association between O 3 and OC found in our study is consistent with previous studies ( Gilboa et al., 2005 ; Hwang and Jaakkola, 2008 ) . For example , Hwang et al. 2008 found a 17% increase in risk of CLP for a 10 ppb increase in O 3 exposure during the first month of pregnancy among Taiwanese newborns during 2001 2003 ( Hwang and Jaakkola, 2008 ) . O 3 exposure has been found to be associated with decreased immune functions, increase d systemic infl ammation , and epithelial cell deaths ( Kahle et al., 2014 ; Murphy et al., 2014 ; Sharkhuu et al., 2011 ) . These respon ses have bee n suggested to have negative effects on maternal health, maternal fetal exchange, and can subsequently influence the risk of fetal development ( Kannan et al., 2006 ; Slama et al., 2008 ) . Therefore, these mechan isms may partly explain the observed positive associations between O 3 and OC. T he mechanism behind the stronger association among female infants is unclear. There is evidence that females are in general more susceptible to the effects of air pollution and smoking on health ( Brunekreef et al., 1997 ; Kan et al., 2007 ;

PAGE 98

98 Lodrup Carlsen et al., 2006 ; Pershagen et al., 1995 ; Tong et al., 2015 ; van Vliet et al., 1997 ) . However, w e are not aware of any study that had addressed this question for fetal exposure in the literature; therefore, we are unable to compare our findings with existing evidence . Nevertheless, our results may imply differences in the biologic mechanisms underl ying the association between O 3 and OC among males and females . Some studies have suggested that maternal hormonal profile around the time of conception time can partially determine the sex of the infant ( James, 2000 ; James, 2008b ) . Meanwhile, air pollution exposure has been observed to affect endocrine function among pregnant women ( Kannan et al., 2006 ; Slama et al., 2008 ) . A study in Brazil suggested that higher level of air pollution exposure is associated with higher female births ( Mirag lia et al., 2013 ) . Therefore, it is possible that a hormone mediated pathway is underlying the higher sensitivity to air pollution among females. More studies are needed to confirm our findings and to understand the associated biologic mechanisms . Finall y, our findings on the differences in characteristics between cases and controls are consistent with the literature ( Mossey et al., 2009 ) . However, some of the differences were not statistically significant for CPO, perhaps due to small sample size. For gestational complications, CPO cases were more likely to be affected; however, CLP cases were not. The reason for th is discrepancy is not clear; however, we suspect that this may be a result of under reporting of gestational comp lications in Vital Statistics. This study has several limitations. While the HBM model allows better spatial coverage compared to closest air monitor , it only outputs a prediction for each 12x12km

PAGE 99

99 grid on the map. This resolution is not sensitive to variat ion in exposure at a smaller scale , which may lead to a source of potential misclassification . This may also bias our results towards the null when matched cases and controls come from the same area, leading to little exposure contrast. There were 12 cases in which at least one of their the same results. Additionally, the use of residential address at del ivery to assess exposure assumes that the mother s did not relocate thr oughout pregnancy. However, this may not be true for all women , leading to potential misclassification. Unfortunately, since there is no residential mobility information, it is not possible to quantify this potential bias. Studies have shown that up to 30 percent of women change residence between conception and birth ( Canfield et al., 2006b ; Fell et al., 2004 ) . However, a majority of these moves appear to be local, to areas with a similar socioeconomic make up and the characteristics of women who move are similar to those who do not ( Canfield et al., 2006b ) . Furthermore, even if the mother stayed at the same address through pregnancy and birth, her daily activity patterns and time spent at work may lead to another source of potential misclassification. Additionally, we did not have data on maternal occupation, or indoor pollution level, which could have confounded our results. The fact that Vital Statistics only includes birth defects diagnosed at birth could lead to potential exclusion of cases. However, since the outcome of this study is OC , t his may not be an issue as OC are very apparent at birth. Most OC are diagnosed at birth or even before birth by ultrasound ( CDC, 2014 ) . A capture recapture analysis h as shown that Vital Statistics birth certificate data have high completenes s for OC ( Wa ng et

PAGE 100

100 al., 2006 ) . Even if we missed some potential cases and they were included among controls, this would have biased our res ults towards the null, and cannot explain the positive association between air pollutants and OC found in this study. Furthermore, one may question the reliability of maternal smoking inf ormation on Vital Statistics . A recent reliability study of maternal smoking on Vital Statistics birth data from several states have suggested that it has exceptionally high agreement with both maternal worksheets (self reported) and medical records ( Howland et al., 2015 ) . However, if maternal smoking is under reported in birth certifica te, and this is non differential among cases and controls, then this could have biased our results toward null differences between th e smoking and no smoking group. Since our study has four different windows of exposures, our significant findings for effec t modification in some exposure windows may have been spurious due to multiple testing. L astly, selection bias may also be an issue of this study. In order to assess whether this is an issue, the study sample was also compared to the source population for differences in important socio demographic characteristics. The results suggested that they were comparable , suggesting no substantial sele ction bias (results not shown). Conclusion This study found that exposures to PM 2.5 or O 3 may increase the risk of OC . However, these effects varied by types of OC, exposure period, maternal smoking status , and infant sex. Given the public health importance of OC and the ubiquitous nature these air pollutants, it is important for continued efforts to raise awareness on the association between air pollutants on OC , reduce air pollutant exposures, and further

PAGE 101

101 understand the sources and the mechanisms through which certain air pollutants may increase the risk OC. Meanwhile more studies are needed to confirm our findings

PAGE 102

102 Table 4 1. Characteristics of study participants in Florida, 2004 2008 Characteristics CLP CPO Cases (n=478) Controls (n=1,912) p value Cases (n=153) Controls (n=612) p value N % N % N % N % Infant sex 0.0002 0.0467 Female 193 40.4 954 49.9 88 57.5 297 48.5 Male 285 59.6 958 50.1 65 42.5 315 51.5 Maternal race <.0001 0.0012 White 283 59.2 869 45.7 95 62.1 272 44.4 Black 39 8.2 323 17.0 11 7.2 97 15. 9 Hispanic 143 29.9 626 33.0 40 26.1 216 35. 3 Other 13 2.7 82 4.3 7 4.6 25 4. 1 Missing 0 0 0 0 0 0 2 0.3 Maternal age 0.3378 0.8987 <20 58 12.1 211 11.0 15 9.8 57 9.3 20 34 366 76.6 1 , 439 75. 3 121 79.1 479 78. 3 35 54 11.3 262 13.7 17 11.1 76 12.4 Maternal education 0.1216 0.1370 HS 262 54.8 957 50. 1 93 60.8 317 51.8 College 211 44.1 941 49.2 59 38.6 290 47. 4 Missing 5 1.0 14 0.7 1 0. 7 5 0.8 Marital status 0.1418 0.2551 Married 261 54.6 1115 58.3 96 62.7 353 57. 7 Not married 217 45.4 797 41. 7 57 37.3 259 42.3 Parity 0.1380 0.8837 0 196 41.0 856 44. 8 64 41.8 260 42. 5 282 59.0 1 , 056 55.2 89 58.2 352 57.5 Smoking 0.0002 0.2956 Yes 68 14.2 165 8.6 18 11.8 55 9.0 No 410 85.8 1 , 747 91. 4 135 88.2 557 91.0

PAGE 103

103 Table 4 1. Continued Characteristics CLP CPO Cases (n=478) Controls (n=1,912) p value Cases (n=153) Controls (n=612) p value N % N % N % N % Alcohol 0.8246 1.0000 Yes 2 0.4 5 0.3 1 0.7 4 0.7 No 476 99.5 1,907 99.7 152 99.4 608 99.3 Pre pregnancy BMI 0.1089 0.7589 <18.5 57 11.9 215 11.2 18 11.8 78 12.8 18.5 24.9 217 45.4 904 47.3 69 45.1 283 46.2 25 29.9 102 21.3 468 24.5 32 20.9 138 22.6 102 21.3 325 17.0 34 22.2 113 18.5 Gestational complications 0.7346 0.0283 Yes 40 8.4 142 7.4 20 13. 1 43 7.0 No 437 91.4 437 92.3 132 86. 3 568 92.8 Missing 1 0.2 6 0.3 1 0. 7 1 0.2 Neighborhood income 0.1200 0.5388
PAGE 104

104 Table 4 2. Estimated exposure to ambient air pollutants at birth residence for cases and controls in Florida, 2004 2008 . Pollutants and exposure windows CLP CPO Cases Controls Cases Controls Mean SD Mean SD Mean SD Mean SD PM 2.5 3 ) Week 1 2 9.6 3.2 9. 6 3. 0 9.2 3.0 9. 3 3.2 Week 3 8 9. 7 2.8 9.6 2. 8 9.6 3.2 9.7 3.3 Week 9 13 9. 5 2.6 9.6 2.7 9.6 2.7 9. 8 2.8 Trimester 1 9. 7 2.0 9.6 2.1 9.5 2.2 9.7 2. 4 O 3 (ppb) Week 1 2 38.8 7.4 37.7 7.6 39.2 7.8 38.0 7.7 Week 3 8 40.0 7.2 37.8 7.4 39.6 7.7 38. 0 7.7 Week 9 13 38.7 7.2 38.0 7.4 39.2 6.7 38.2 7.3 Trimester 1 38.8 6.0 37.9 6.4 39.3 6.2 38.1 6.4 Abbreviations: CLP, cleft lip with or without cleft palate; CPO, cleft palate only; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 m icron s ; SD; standard deviation.

PAGE 105

105 Table 4 3. Associations between PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy among Florida births, 2004 2008 . PM 2.5 3 ) Unadjusted OR a (95% CI) Adjusted ab OR (95% CI) Overall Smoking No smoking Overall Smoking No smoking CLP Week 1 2 1.00 (0.96,1.04) 1.06 (0.96,1.17) 1.03 (0.97,1.08) 1.00 (0.96,1.05) 1.06 (0.96,1.17) 1.03 (0.97,1.09) Week 3 8 1.04 (1.01,1.07) * 1.12 (1.01,1.24) 1.05 (1.00,1.11 ) 1.05 (1.00,1.09) * 1.11 (1.01,1.25) 1.06 (1.00,1.13) Week 9 13 0.97 (0.92,1.01) 1.00 (0.89,1.10) 0.95 (0.89,1.01) 0.97 (0.91,1.03) 1.00 (0.90,1.11) 0.96 (0.90,1.03) Trimester 1 1.04 (0.99,1.10) * 1.11 (0.97,1.28) 1.02 (0.94,1.11) 1.05 (0.99,1.10) * 1.12 (0.97,1.30) 1.04 (0.95,1.13) CPO Week 1 2 0.92 (0.84,1.01) 0.94 (0.78,1.13) 0.92 (0.82,1.03) 0.91 (0.82,1.01) 0.98 (0.79,1.20) 0.94 (0.83,1.06) Week 3 8 0.98 (0.91,1.05) 0.90 (0.74,1.11) 0.95 (0.85,1.06) 1.00 (0.92,1.07) 0.89 (0.71,1.12) 0.94 (0.83,1.07) Week 9 13 0.93 (0.86,1.02) 0.86 (0.68,1.08) 0.90 (0.79,1.02) 0.94 (0.87,1.04) 0.88 (0.68,1.15) 0.92 (0.79,1.06) Trimester 1 0.90 (0.79,1.02) 0.82 (0.61,1.11) 0.87 (0.73,1.03) 0.93 (0.84,1.03) 0.86 (0.62,1.19) 0.89 (0.74,1.07) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less tha n 2.5 micron; OR, odds ratio. a Odds ratios are for a 3 increase in PM 2.5 concentration. b Models were adjusted for date of birth, infant sex, maternal age, maternal race, mate rnal education, marital status , gestational age, gestational complications, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall estimates were also adjusted for maternal smoking during preg nancy. *indicates statistically significant interaction term at p<0.05.

PAGE 106

106 Table 4 4. Associations between O 3 and oral clefts by exposure windows and infant sex among Florida births, 2004 2008 . O 3 ( 1ppb) Unadjusted OR a (95% CI) Adjusted OR ab (95% CI) Overall Female Male Overall Female Male CLP Week 1 2 1.05 (1.02,1.07) * 1.06 (1.03,1.09) 1.03 (1.01,1.06) 1.03(1.01,1.06) 1.05 (1.01,1.08) 1.02 (0.99,1.05) Week 3 8 1.06 (1.04,1.08) * 1.08 (1.05,1.11) 1.04 (1.02,1.07) 1.04(1.01,1.06) * 1.07 (1.03,1.10) 1.03 (1.01,1.06) Week 9 13 1.04 (1.02,1.06) * 1.06 (1.03,1.09) 1.02 (0.99,1.05) 1.02(0.99,1.05) * 1.04 (1.01,1.08) 1.00 (0.97,1.03) Trimester 1 1.06 (1.03,1.09) * 1.09 (1.05,1.13) 1.04 (1.01,1.07) 1.04(1.01,1.07) * 1.07 (1.03,1.11) 1.02 (0.99,1.05) CPO Week 1 2 1.08 (1.04,1.13) 1.09 (1.04,1.14) 1.09 (1.03,1.14) 1.07 (1.02,1.12) 1.06 (1.01,1.12) 1.07 (1.01,1.14) Week 3 8 1.09 (1.05,1.14) 1.09 (1.04,1.14) 1.10 (1.05,1.16) 1.07 (1.02,1.13) 1.07 (1.01,1.13) 1.08 (1.02,1.14) Week 9 13 1.07 (1.02,1.11) 1.07 (1.02,1.12) 1.06 (1.01,1.11) 1.04 (0.99,1.10) 1.04 (0.99,1.10) 1.04 (0.98,1.10) Trimester 1 1.10 (1.05,1.15) 1.10 (1.04,1.16) 1.11 (1.04,1.17) 1.08 (1.02,1.14) 1.07 (1.01,1.14) 1.08 (1.01,1.16) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 micron; OR, odds ratio. a Odds ratios are for a 1 ppb increase in O 3 concentration. b Models w ere adjusted for date of birth, maternal age, maternal race, mate rnal education, marital status , maternal smoking during pregnancy, gestational age, gestational complications, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall estimates were also adjusted for infant sex. *indicates statistically significant interaction term at p<0.05. .

PAGE 107

107 Table 4 5. Associations between PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy in a multi pollutant model, 2004 2008 . PM 2.5 ( 3 ) Adjusted ab OR (95% CI) Overall Smoking No smoking CLP Week 1 2 1.02 (0.98,1.06) 1.0 5 (0.9 5 , 1.1 6 ) 1.0 1 (0.9 5 , 1.0 7 ) Week 3 8 1.05 (0.99,1.10) * 1. 09 ( 0.98 , 1.2 2 ) 1.0 4 ( 0.97 , 1.1 1 ) Week 9 13 1.00 (0.95,1.05) 0.98 (0.89, 1.10 ) 0.95 (0.89 , 1 .02 ) Trimester 1 1.03 (0.98,1.08) * 1.1 1 (0.96, 1.29 ) 1.02 (0.94, 1.11 ) CPO Week 1 2 0.92 (0.82,1.01) 0.9 1 (0.7 5 ,1. 12 ) 0. 89 (0.78,1.01 ) Week 3 8 0.95 (0.87,1.03) 0.8 6 (0.68,1.09 ) 0.90 (0.80,1.03 ) Week 9 13 0.92 (0.84,1.01) 0.8 5 (0.66,1.09 ) 0.89 (0.7 7 ,1.0 2 ) Trimester 1 0.90 (0.81,1.01) 0.8 1 (0. 58,1.11 ) 0.85 (0.71 ,1.0 2 ) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 micron; OR, odds ratio. a Odds ratios are for a 3 increase in PM 2.5 concentration. b Models were adjusted for date of birth, infant sex, maternal age, maternal race, mate rnal education, marital status , gestational age, gestational complications, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall estimates were also adjusted for maternal smoking during pregnancy. *indicates statistically significant interaction term at p<0.05.

PAGE 108

108 Table 4 6 . Associations between O 3 and oral clefts by exposure windows and infant sex in a multi pollutant model, 2004 2008 . O 3 ( 1ppb) Adjusted OR ab (95% CI) Overall Female Male CLP Week 1 2 1.04(1.01,1.07) 1.05 (1.02,1.09) 1.03 (1.00,1.06) Week 3 8 1.04(1.01,1.07) * 1.06 (1.02,1.09) 1.0 2 (1.00,1.0 5 ) Week 9 13 1.03(0.99,1.06) * 1.05 (1.02,1.09) 1.01 (0.98,1.04) Trimester 1 1.04(1.01,1.08) * 1.07 (1.03,1.11) 1.02 (0.99,1.06) CPO Week 1 2 1.10(1.04,1.16) 1. 10 (1.0 3 ,1.1 7 ) 1.0 9 (1.0 3 ,1.1 6 ) Week 3 8 1.09(1.03,1.15) 1.0 9 (1.0 2 ,1.1 6 ) 1.08 (1.02,1.1 5 ) Week 9 13 1.05(1.00,1.10) 1.0 5 (0.99,1.1 1 ) 1.04 (0.98,1.1 1 ) Trimester 1 1.09(1.03,1.16) 1.0 9 (1.01,1.1 7 ) 1.0 9 (1.01,1.1 7 ) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 micron s ; OR, odds ratio. a Odds ratios are for a 1 ppb increase in O 3 concentration. b Models were adjusted for date of birth, maternal age, maternal race, mate rnal education, marital status , maternal smoking during pregnancy, gestational age, gestational complications, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall estimates were also adjusted for infant sex. *indicates statistically significant interaction term at p<0.05. .

PAGE 109

109 Table 4 7 . Associations between nearest monitor PM 2.5 and oral clefts by exposure windows and maternal smoking during pregnancy among Florida births, 2004 2008 . PM 2.5 ( 3 ) Adjusted ab OR (95% CI) Overall Smoking No smoking CLP Week 1 2 1.01(0.95,1.07) 1.05 ( 0.98 , 1.11 ) 1.02 ( 0.97 , 1.08 ) Week 3 8 1.02(0.96,1.08) * 1.06 ( 1.01 , 1.13 ) 1.03 ( 0.97 , 1.10 ) Week 9 13 1.00(0.93,1.06) 1.02 ( 0.95 , 1. 09 ) 0.99 ( 0.93 , 1.06 ) Trimester 1 1.03(0.95,1.11) * 1.08 (0.91, 1.27) 1.02 (0.92,1.12) CPO Week 1 2 1.04 (0.96,1.12) 1.08 ( 0.97 , 1.20 ) 1.04 ( 0.95 , 1.13 ) Week 3 8 1.06 (0.95,1.15) 1.08 ( 0.96 , 1.20 ) 1.04 ( 0.95 , 1.15 ) Week 9 13 1.02 (0.92,1.13) 1.05 ( 0.93 , 1.20 ) 1.02 ( 0.91 , 1.14 ) Trimester 1 1.05 (0.94,1.16) 1.08 ( 0.94 , 1.24 ) 1.04 ( 0.92 , 1.18 ) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 micron s ; OR, odds ratio. a Odds ratios are for a 3 increase in PM 2.5 concentration. b Models were adjusted for date of birth, infant sex, maternal age, maternal race, mate rnal education, marital status , gestational age, gestational complications, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall e stimates were also adjusted for maternal smoking during pregnancy. *indicates statistically significant interaction term at p<0.05. .

PAGE 110

110 Table 4 8 . Associations between nearest monitor O 3 and oral clefts by exposure windows and infant sex among Florida births, 2004 2008 O 3 ( 1ppb) Adjusted OR ab (95% CI) Overall Female Male CLP Week 1 2 1.04 (1.02,1.06) * 1.06 (1.02, 1.10) 1.03 (1.00, 1.06) Week 3 8 1.0 4 (1.01,1.06) 1.05 (1.01, 1.09) 1.04 (1.01, 1.07) Week 9 13 1.00 (0.99,1.03) 1.03 (0.99, 1.07) 1.00 (0.97, 1.03) Trimester 1 1.03 (1.00,1.06) * 1.06 (1.01, 1.11) 1.03 (0.99, 1.06) CPO Week 1 2 1.04 (0.99,1.08) 1.04 (1.00,1.08) 1.03 (0.97,1.10) Week 3 8 1.06 (1.01,1.11) 1.06 (1.01,1.11) 1.06 (0.99,1.13) Week 9 13 1.05 (1.00,1.10) 1.05 (1.00,1.10) 1.05 (0.98,1.13) Trimester 1 1.07 (1.01,1.13) 1.07 (1.01,1.14) 1.09 (0.99,1.19) Abbreviations: CLP, cleft lip with or without palate; CPO, cleft palate only ; O 3 , ozone; PM 2.5 , particulate matter with diameter less than 2.5 micron s ; OR, odds ratio. a Odds ratios are for a 1 ppb increase in O 3 concentration. b Models w ere adjusted for date of birth, maternal age, maternal race, mate rnal education, marital status , maternal smoking during pregnancy, gestational age, gestational complic ations, pre pregnancy BMI, parity, neighborhood income, urbanity , maternal country of origin . Overall estimates were also adjusted for infant sex. *indicates statistically significant interaction term at p<0.05. .

PAGE 111

111 Figure 4 1. Distribution of active air monitors in Florida during the study period

PAGE 112

112 CHAPTER 5 CONCLUSIONS Summary of Research Objectives Following the theoretical framework outlined in Chapter 1 , aside from individual risk factors, the environment may also independently influence the risk of OC . This dissertation aimed to contribute to the limited literature on OC etiolog y by exploring e nvironmental determinants and identifying potential high risk or susceptible subgroups . In order to achieve these objective s, we first systematically review ed and perform ed a meta analysis on population based studies evaluating the relationship between criteria air pollutants and OC outcomes to determine if there is sufficient evi dence to suggest an association . Second , we use d spatial scan statistics to identify potential OC cluster(s) in Florida from 2004 2008 . We then compare d characteristics (e.g. annual air pollution concentrations and SES) of neighborhoods inside versus those outside of the cluster s . Since the second aim was limited from ecological fallacy and did not account for t ime of exposure to air pollution , we further conduct ed a population based case control study to determine the association of OC with exposures to PM 2.5 and O 3 during the first trimester of pregna ncy . Due to the lack of literature on susceptibl e population s , we also investigated whether maternal and infant characteristics including maternal race, maternal smoking during pregnancy, and infant sex modify the associations between the two air pollutants and OC . More specifically, we sought to answer the following research questions: 1. What are the overall effect estimates for the association between criteria pollutants and OC based on existing literature? 2. Were there any clusters of high OC r ates in the state of Florida from 2004 2008 after adjusting for individual factors?

PAGE 113

113 3. Which neighborhood level characteristics were associated with high risk OC clustering? 4. Were there any associations between PM 2.5 or O 3 expo sures during first trimester and OC; and, if so, what was the magnitude of these associations ? 5. Were these associations modified by individual risk factors including infant sex , maternal smoking status , and maternal race ? Accomplishments of this Dissertation Taken together, this dissertati on has accomplished the outlined objectives and successfully answered the research questions outlined in Chapter 1 . T he findings have addressed several gaps in the current literature pertaining to the environmental risk factors of OC. These achievements ar e summarized and discussed in the following sections. In addition, a summary of findings are presented in Figure 5 1. Meta analysis of Relevant Literature Due to recent concerns regarding the potential association between air pollution and adverse birth outcomes, there has been an increase in the number of studies evaluating the associations between c riteria air pollutants and OC. However, g iven the differences in designs across studies, it is difficult to draw a firm conclusion. Therefore, in Chapter 2 we conducted a systematic review and meta analysis on the existing literature investigating the association s between criteria air pollutants and OC . To answer the first research question , we found evidence for association between some criteri a air pollutants and the risk of OC. Specifically, w e found that SO 2 exposure during weeks 3 8 of gestation was signif icantly associated with the risk of CLP. On the other hand, CO exposure during the first trimester had an inverse association with CPO. Ou r findings add to the literature on this topic the summary estimates of the association between different criteria air pollut ants and OC. I n addition, t he identification of a

PAGE 114

114 potential susceptible window of exposures for OC in this analysis (weeks 3 8 of g estation) has helped form additional research questions and hypotheses that were explored in Chapter 4 of this dissertation. This study also identified some current methodological shortcomings in the field , some of which we addressed in subsequent parts o f the dissertation . Some of these limitations include 1 ) dependence on closest air monitors for air pollution assessment and 2 ) lack of identification of susceptible subgroups. Additionally, we noticed that although PM 2.5 and O 3 have been the most implicated air pollutants for health, there is a lack of consistent evidence of association with OC. Therefore, in subsequent chapters, we focused on these two air pollutants. Identification of Oral Cleft s Clusters in Florida Since risk factors for OC can be unequally distributed through geographic location, there is a potential that OC may also spatially cluster. Thus , in order to further explore environmental risk factors for OC , Chapter 3 presents a spatial cluster analysis using spat ial scan statistic to determine if there was significant OC cluster(s) after adjusting for significant individual risk factors. To answer the secon d research question , w e found two spatial clusters of OC in Florida from 2004 20 08 after adjustment for impor tant individual covariates including infant sex , gestational complications, maternal education, maternal country of origin, maternal race, and maternal smoking during pregnancy. The identification of these two potential OC clusters with significantly eleva ted risk compared to expectation is of great public health interest. Although the clusters did not contain the majority of the cases in Florida, they suggest that OC occur red at much higher rates than expected in these two regions even after adjustment for known individual risk fa ctors. Going back to the theoretical framework, this finding

PAGE 115

115 may suggest that individuals living inside the clusters may have been disproportionally exposed to some comm on environmental risk factors that may have contributed to their increased risk . Therefore, the identification of these spatial clusters may potentially help future studies to generate hypothesis and reveal additional risk factors. Furthermore, our results may be relevant for identification of high risk neighborho ods to aid with health services planning, allocation of resources, and other prevention and management efforts. This analysis also highlight ed the usefulness and feasibility of spatial scan statistics in studying birth outcomes which have relatively short latency period. Oral clefts are particularly well suited to this type of geographic analyses because the lag time between environmental exposures and the development of birth outcome s is relativel y short. This minimizes the potential bias introduced when study participants relocate d uring the exposure period and allows for stronger hypotheses about the ar ea level factors that may increase the risk of OC. Determination of Neighborhood Predictors of Oral Clefts Clustering After we identified clusters of OC i n which the rates were si gnificantly higher than the rest of the state, we investigated the neighborhood characteristic s that may be associated with OC clustering . To answer the third research question, at the neighborhood (e.g. census block group (CBG) ) level, we found positive association s between OC clustering and average annual concentrations of PM 2.5 and O 3 as well as residence in rural areas . We found no evidence of association between OC clustering and neighborhood SES. As shown in Chapter 2, the as sociations of OC with PM 2.5 and O 3 are still inconsistent at the individual level in the literature . Therefore, th ese ecologic finding s suggest a potential association between the se two air pollutants and OC that

PAGE 116

116 needs further investigation in Florida . The fact that rurality showed strong association with OC clustering may suggest that pregnant women in rural areas may have different exposures compared to those in urban areas. Since Florida has high agricultural land coverage, agriculture related risk facto rs may be implied ( Engel et al., 2000 ; Heeren et al., 2003 ; Langlois et al., 2009b ; Ochoa Acuna and Carbajo, 2009 ) . In addition, women in rural areas may have limited access to certain health related amenities and resources that may affect their risk ( Hartley et al., 1994 ; Slifkin et al., 2004 ) . Due to the ecolog ic nature of this study, we conduct ed a more refined case control study to determine prenatal exposures to PM 2.5 and O 3 and their association s with OC in Chapter 4. Determination of Association of Air Pollution with Oral Clefts an d Identification of Susceptible Subgroups To build on previous findings, we conducted a population based case control study to determine the associations of prenatal exposures to PM 2.5 as well as O 3 with OC. To answer the fourth and fifth research question, we found that PM 2.5 exposures during weeks 3 8 of gestation were positively associated with the risk of CLP among infants of mothers wh o smoked during pregnancy ; however, the associations among those unexposed to maternal smoking during pregnancy was not significant . On the other hand, we found no association between PM 2.5 and CPO. Our findings add important contribution to the literature s ince this is one of the first to report that PM 2.5 may have association with OC among infants of w omen who smoke during pregnancy . Future studies can confirm these findings and investigate the mechanisms underlying the interaction between smoking and PM 2.5 . Our findings also confirm the critical exposure window for PM 2.5 during weeks 3 8 of pregnancy. We also found that O 3 exposures during the first trimester were positively associated with the risk of OC. This

PAGE 117

117 is the first study to report that female infants were potentially more susceptible to the effects of O 3 on CLP . This can suggest differences in the biologic mechanisms linking air pollution and OC between the two sex groups. To sum up our findings, we found evidence that the environment, particularly air pollution exposure , may have independent associations with the risk of OC after adjustment for known individual risk factors. Secondly, we also found evidence that some geographic locations had higher OC risk compared to others, and that this elevated risk was associated with neighborhood level average annual concentration of PM 2.5 and O 3 , and rura lity . Thirdly, we confirmed that PM 2.5 and O 3 exposures may have positive associations with OC at the individual level, and that these associations may vary by windows of exposure, infant sex, and maternal smoking status. More research is needed to underst and the associations observed above, and mechanisms of interaction between these risk factors. Biological Mechanisms Linking Air Pollution and Oral Clef ts The causal biologic pat hways linking air pollution and OC or any other adverse birth outcomes are currently un clear . However, w ith advances in biomolecular techniques, there is a recent increase in the number of studies showing evidence towards potential mechanisms linking air pollution with adverse birth outcomes ( Kannan et al., 2006 ; Slama et al., 2008 ) . These studies suggest that the underlying mechanisms may involve several non exclusive pathways including oxidative stress and inflammation; alteration of endothelial, rheologic , and hemodynamic function s; endocrine disruption; and epigenetic and genetic changes ( Kannan et al., 2006 ) . Figure 5 2 illustrates these mechanisms.

PAGE 118

118 Oxidative S tress and Inflammation Oxidative stress is a condition in which the body encounters more reactive oxygen or nitrogen species than its ability to properly remove them, resulting in excess peroxides and free radicals that can ultima tely damage cell components and/or biological processes ( Mittal et al., 2013 ) . Exposure to air pollution has been consi stently shown to increase the concentration of oxidative stress markers ( Ghio et al., 2012 ; Nagiah et al., 2014 ) . For example, PM can activate inflammatory cells to generate oxidative stressors such as reactive oxygen species and reactive nitrogen species ( Tao et al., 2003 ) . Consequently, pollution induced inflammation can follow as a direct result of the oxidation of these reactive species, or by up regulation of pro inflammatory mediators such as cytokines and chemokines directly induced by air pollutants ( Risom et al., 2005 ; Sioutas et al., 2005 ) . These inflammatory processes have been sugge sted to affect maternal immunity, leading to increased risk of infection, which ultimately may lead to adverse birth outcomes ( Slama et al., 2008 ) . Furthermore, the release of cytokines and chemokines following exposure to pollution can also cause acute placental inflammation ( Becker et al., 2005 ) . This process can subsequently result in cell apoptosis at the placenta and ultimately impair trans place ntal nutrient exchanges ( Sharp et al., 2010 ) . In addit ion, there is also evidence from animal experiment as well as human studies suggesting that air pollution induced oxidative stress can lead to DNA damage ( Risom et al., 2005 ; Whyatt et al., 1998 ) , which can ultimately affect fetal development and cause growth restriction and low birth weight ( Rossner et al., 2011 ) . Hartwig et al. 2002 suggested that oxidative stress can interfere with DNA repairing mechanisms durin g transcription, leading to higher rates of DNA damage ( Hartwig et al., 2002 ) . If

PAGE 119

119 the fetus during development, they all may affect fetal development, thus leading to adverse birth outcomes. Alteration of Endothelial, R heo logic and Hemodynamic Functions In order for the fetus to develop properly, there needs to be sufficient umbilical b lood flow and maternal fetal exchange of nutrients, oxygen, and other vital substances. Alterations of this exchange process can severely affect the developing fetus ( Pardi et al., 2002 ) . Factors that can adversely affect the proper nutrient and oxygen transport include endothelial functions (e.g. vaso constriction activities), rheologic factors (e.g. blood viscosity and coag g ulability ), and hemodynamic factors (e.g. blood pressure . T hese functions have been found to be affected by exposure to PM among the general population ( Krishnan et al., 2012 ; Pekkanen et al., 2000 ; Peters et al., 1997 ; Pope and Dockery, 2006 ) . Specifically, exposure to fine PM has been shown to increase blood viscosity and co agulation, decrease blood flow, and increase vasoconstriction activities ( Krishnan et al., 2012 ; Peters et al., 1997 ; Pope and Dockery, 2006 ) . Although no studies have evaluated these res ponses among pregnant women, who in general may have different vascular factors compared to the general population, it is possible that the responses are similar among this population . If so, pollution induced changes in blood viscosity and endothelial cha nges may interfere with maternal fetal exchanges by decreasing placental blood flow and nutrient supply , which may ultimately affect fetal development ( Proietti et al., 2013 ) . In addition, exposures to air pollutants such as PM 2.5 and O 3 have also been found to change hemodynamic balance by incre asing blood pressure during pregnancy ( Lee et al., 2012 ; Mobasher et al., 2013 ) . It has also been suggested that air pollution

PAGE 120

120 can increase sympathetic tone ( Ibald Mulli et al., 2001 ) . In addition, exposure can also inhibit the action of nitrous oxide and in turn prevent vasodilation ( Delfino et al., 2005 ) . These effects may lead to increase in blood pressure and hypertension. Since hypertension durin g pregnancy is a risk factor for fetal malnutrition after controlling for important confounders ( Deodhar and Jarad, 1999 ) , there may be an underlying mechanism through which hypertension may be associated with alter ed mat ernal fetal nutrient exchange. Moreover, there is also evidence that women with hypertensive disorders of pregnancy such as preeclampsia have lower levels of placenta growth factor, which is used to predict risk of adverse birth outcomes in clinical practice ( Levine et al., 2004 ) . Placenta growth factor is important for the development of the p lacenta; therefore, lower levels may suggest insufficient placental growth. This may lead to insu fficient maternal fetal nutrient/oxygen exchange, which can negatively affect the fetus. This mechanism becomes more plausible given the numerous studies suggesting the strong association between gestational hypertension and adverse pregnancy outcomes incl uding OC in some cases ( Lebby et al., 2010 ; Orbach et al., 2013 ) . Given the evidence that hypertension during pregnancy affects placental gro wth and increases risk for fetal malnutrition, it is likely that air pollution may potentially be associated with adverse birth outcomes including birth defects through altering maternal fetal exchange via this pathway. Endocrine Disruption Although this pathway has been less studied, there is evidence linking air pollution exposure to endocrine dysfunctions, which may have a negative impact on birth outcomes. For example, PM in diesel exhaust has been shown to induce changes in estrogenic activities and p rogesterone production in the reproductive system both in

PAGE 121

121 vivo and in vitro ( Furuta et al., 2004 ; Takeda et al., 2004 ; Tomei et al., 2007 ; Tra n et al., 1996 ) . The components of PM may mimic or partially mimic naturally occurring hormones such as estrogen or androgen , thus may bind to the hormone receptors and block the endogenous hormone from binding. The normal signal then fails to occur and the body fails to respond properly . This may also induce an over production of the endogenous hormones. There is also evidence suggesting that environmental tobacco smoke (ETS), which has many similar ities in characteristics with PM air pollution, can contribute to thyroid dysfunction. Specifically, ETS can induce the release of oxidative markers within the thyroid, impair thyroi d function and iodine uptake, and increase thyroid autoimmunity ( Carrillo et al., 2009 ) . These activities may lead to initiation of catabolic metabolic processes and higher resting energy expenditure up to 10% ( Carrillo et al., 2009 ) . Disruption of thyroid function has been associated with improper fetal development and other adverse pregnancy outcomes in previous studies ( Sarkar, 2012 ; Su et al., 2011a ; van den Boogaard et al., 2011 ) . The changes in estrogenic activities and endocrine disruption resulting from air pollution exposures can also directly affect epidermal growth factor and insulin like growth factors types I a nd II receptors , leading to inhibition of placental cell growth and proliferation ( Turgut et al., 2005 ) . These effects could subsequently cause decreased maternal fetal exchange and in turn cause restricted fetal growth ( Kan aka Gantenbein et al., 2003 ) ; therefore, the link between air pollution and OC may be speculated to i nvolve this pathway.

PAGE 122

122 Genetic and Epigenetic Changes A nimal studies have demonstrated that germ line mutation rate s are significantly higher in mice and birds that were exposed to more pollution ( Somers et al., 2004 ; Somers et al., 2002 ; Yauk et al., 2008 ) . Air p ollution has also been shown to increase heritable DN A abnormality , and decrease human sperm quality (e.g., abnormal sperm chromatin, head shape, and decreased mobility and concentration) ( Seleva n et al., 2000 ; Somers and Cooper, 2009 ; Somers et al., 2002 ) . Moreover, several studies have also suggested that exposure to air pollution is associated with elevated levels of DNA fragmentation in human spe rms ( Jafarabadi, 2007 ; Rubes et al., 2005 ) . Among pregnant women, exposure to high air pollution levels has also been linked to compr om ised DNA integrity and repair mechanisms ( Nagiah et al., 2014 ) . Given its ability to affect heritable DNA and interfere may be involved in the association with adverse birth outcomes through this mechanism. More recently, PM and O 3 have also been shown to have the potentia l to induce epigenetic changes including DNA methylation and histo ne modification. They can change the DNA methylation level related to inflammation and oxidative stress marker expression ( Baccarelli et al., 2009 ; Barthauer, 1990 ; Fry et al., 2014 ; Janssen et al., 2013 ; Ji and Khurana Hershey, 2012 ; Miousse et al., 2014 ) . These epigenetic activities have also been observed in the placenta during early pregnancy ( Janssen et al., 2013 ; Janssen et al., 2012 ; Koukoura et al., 2012 ) and may be part of etiologies of fetal growth and birth def ect s ( Banister et al., 2011 ) . Future D irections One of the major limitations o f the current study as well as the existing literature involves potential lack of spatial and temporal accuracy of air pollution data. Due to the

PAGE 123

123 cost of personal monitoring, many s tudies still rely on the closest monitor station, or more recently, geostat is ti cal models to estimate exposures ( Farhi et al., 2014 ; Gilboa et al., 2005 ; Hansen et al., 2009 ; Hwang and Jaakkola, 2008 ; Schembari et al., 2014 ; Vinikoor Imler et al., 2013 ) . Since the estimates are not true personal exposures, misclassification is likely. Therefore, there is a great need to use a prospective stu dy design with personal monitoring to minimize potential misclassification of exposure. This would allow researchers to accurately study the concentration, time, and dose , which would lead to better cumulative and time specific (e.g. gestational window) ex posure measurements. However, this method may not be feasible due to the rarity of OC ( Parker et al., 2010 ) . A more practical alternative would be to develop more spatially and temporally sensitive geostatistical model s that can accurately predict air pollution concentrations at finer scales. I t is also possible that the composition of air pollution is different across geographic areas . These different elements may also have independent effects and/or interact with each other to produce different effects on different neighborhoods or groups of people. Some recent studies have started to investigate the health effe cts of constituents of PM , which is comprised of many different fine metal particles and organic compounds ( Beelen et al., 2014 ; Cassee et al., 2013 ; Wu et al., 2012 ) . However, no study of this nature has been conducted for birth defects . Air pollution research on other health outcomes has identified that certain subgroups of the population may be at higher risk ( Colais et al., 2012 ; Goldberg et al., 2000 ) . However, for the association between air pollution and birth defects, very few studies hav e attempted to identify susceptible subpopulations. This dissertation made

PAGE 124

124 some of the first efforts to identify such population s with respect to infant gender and maternal smoking d uring pregnancy. However, due to lack of data, we were unable to study whe ther other factors such as antioxidant intake can modify the associations between air pollution and OC. Therefore, future studies should focus more on factors that may alleviate the associations between air pollution and adverse health outcomes. The identi fication of such factors is important for public health interventions. While exposures to air pollution have positive associations with birth defects and OC , genetics also has an important influence . Therefore, t he gene environment interaction is vital and should be investigated in future studies. Furthermore, it is also important to design studies to understand the biologic mechanisms that are responsible for the associations between air pollution and birth defects . In terms of social determinants of O C, given the lack of literature, more studies are needed to further understand the role of the contextual environment . In addition, it is also important to determine if neighborhood level contextual factors interact with individual risk factor. For example , it is possible that high risk individuals may live in low risk environment s , and vice versa. Therefore, future studies can also investigate the interaction between individual and contextual characteristics. Lastly, although the field has yet to confirm a causal link between air pollution and OC, simple intervention based studies may contribute to the understanding of this association. For example, since oxidative stress is a likely biologic mecha nism underlying the association between air pollution and bi rth defects, antioxidant intake may provide a reasonable protection against the potential effects of air pollution exposures. A study on fetal health outcomes suggeste d that the positive association

PAGE 125

125 between air pollution and fetal mental development was on ly significant among women who had low antioxidant intake during pregnancy ( Guxens et al., 2012 ) . Studies on the associations between air pollution and other health outcomes such asthma has shown potential beneficial effects of antioxidant intake ( Sienra Monge et al., 2004 ; Su et al., 2013 ) . Thus, an antioxidant based intervention would allow a unique opportunity to further explore the oxidative stress pathway. Meanwhile, efforts that involve educating wome n of child bearing age on the importance of nutrition and other health behaviors should continue to be implemented. Currently, community awareness of potential a dverse effects of air pollution is still not optimal ( Bart on Laws et al., 2015 ) . In addition, very few people change their behavior after learning that air pollution levels are high in their area ( Semenza et al., 2008 ; Wen et al., 2009 ) . Therefore, more efforts are needed with respect to awareness . At the same time, ways to alert the community when the air pollution level is hazardous should be more effectively implemented. Currently, the US EPA has a real time prediction of air pollution levels at any zip code on th eir we bsite; however, in order to access this, one must go through the EPA website. This information is also available on weather websites , but they are not normally available on highly utilized media. In the future, this information would have better impact if it w as accessible through daily weather forecast on television for smart phone applications. This rea dily available inf ormation could prompt women to avoid exposures more effectively . Ways of avoiding exposures that have been suggested in the lite rature are presented in Table 5 1 . O ne must be careful of the unintended consequences of each course of action ( Laumbach et al., 2015 ) . For example, staying indoor s to avoid air pollution exposure may reduce physical activity,

PAGE 126

126 which is beneficial for other health outcomes . Currently, there is no consensus on which method is the most effective.

PAGE 127

127 Table 5 1. Methods to reduce personal exposure to outdoor air pollution Methods Assoc iated tasks Staying indoors Stay indoors Close window Use of air conditioning system Cleaning indoor air Use of air filters Reducing effective inhaled dose Avoid exertion to reduce inhalation rate Avoid near sources Avoid exposures to traffic, sources of combustion, or other point sources Use of personal protective equipments Use masks or respirator Avoid outdoors during time of high concentrations Avoid going outdoors when the predicted concentration is high outdoors

PAGE 128

128 Figure 5 1. Summary o f findings

PAGE 129

129 Figure 5 2 . Plausible biological mechanisms linking air pollution and oral cleft s

PAGE 130

130 LIST OF REFERENCES Abramowicz, S., et al., 2003. Demographic and prenatal factors of patients with cleft lip and cleft palate. A pilot study. J Am Dent Assoc. 134 , 1371 6. Acuna Gonzalez, G., et al., 2011. Family history and socioeconomic risk factors for non syndromic cleft lip and palate: A matched case control study in a less developed country. Biom edica. 31 , 381 91. Agyemang, C., et al., 2009. The effect of neighbourhood income and deprivation on pregnancy outcomes in amsterdam, the netherlands. J Epidemiol Community Health. 63 , 755 60. Alsaad, A. M., et al., 2015. First trimester exposure to topi ramate and the risk of oral clefts in the offspring: A systematic review and meta analysis. Reprod Toxicol. 53 , 45 50. American Lung Association, 2001. Urban air pollution and health inequities: A workshop report. Environ Health Perspect. 109 Suppl 3 , 357 74. Antunes, L. S., et al., 2013. Bmp4 polymorphism is associated with nonsyndromic oral cleft in a brazilian population. Cleft Palate Craniofac J. 50 , 633 8. Antunes, L. S., et al., 2014. The impact of nonsyndromic oral clefts on family quality of life. Spec Care Dentist. 34 , 138 43. Baccarelli, A., et al., 2009. Rapid DNA methylation changes after exposure to traffic particles. Am J Respir Crit Care Med. 179 , 572 8. Baird, P. A., et al., 1991. Maternal age and birth defects: A population study. L ancet. 337 , 527 30. Baird, P. A., et al., 1994. Maternal age and oral cleft malformations: Data from a population based series of 576,815 consecutive livebirths. Teratology. 49 , 448 51. Banister, C. E., et al., 2011. Infant growth restriction is associat ed with distinct patterns of DNA methylation in human placentas. Epigenetics. 6 , 920 7. Barthauer, L., 1990. A piece of my mind. Secret pain. JAMA. 264 , 1457. Barton Laws, M., et al., 2015. Gender, ethnicity and environmental risk perception revisited: T he importance of residential location. J Community Health. [Epub ahead of print].

PAGE 131

131 Becker, S., et al., 2005. Seasonal variations in air pollution particle induced inflammatory mediator release and oxidative stress. Environ Health Perspect. 113 , 1032 8. Beelen, R., et al., 2014. Long term exposure to air pollution and cardiovascular mortality: An analysis of 22 european cohorts. Epidemiology. 25 , 368 78. Bell, M. L., et al., 2008. The relationship between air pollution and low birth weight: Effects by mo ther's age, infant sex, co pollutants, and pre term births. Environ Res Lett. 3 , 44003. Bethea, T., et al., 2014. Neighborhood socioeconomic status in relation to cancer mortality in the black women's health study, 1995 2011. Cancer Epidemiol Biomarkers P rev. 23 , 563 564. Bille, C., et al., 2005. Parent's age and the risk of oral clefts. Epidemiology. 16 , 311 6. Blanco Davila, F., 2003. Incidence of cleft lip and palate in the northeast of mexico: A 10 year study. J Craniofac Surg. 14 , 533 7. Blencowe, H., et al., 2010. Folic acid to reduce neonatal mortality from neural tube disorders. Int J Epidemiol. 39 Suppl 1 , i110 21. Blumenshine, P., et al., 2010. Socioeconomic disparities in adverse birth outcomes: A systematic review. Am J Prev Med. 39 , 263 72. Bobak, M., 2000. Outdoor air pollution, low birth weight, and prematurity. Environ Health Perspect. 108 , 173 6. Bonzini, M., et al., 2010. Impact of ambient air pollution on birth outcomes: Systematic review of the current evidences. Med Lav. 101 , 341 6 3. Botto, L. D., et al., 2014. Congenital heart defects after maternal fever. Am J Obstet Gynecol. 210 , 359 e1 359 e11. Boulet, S. L., et al., 2009. Children with orofacial clefts: Health care use and costs among a privately insured population. Public He alth Rep. 124 , 447 53. Boyles, A. L., et al., 2010. Maternal alcohol consumption, alcohol metabolism genes, and the risk of oral clefts: A population based case control study in norway, 1996 2001. Am J Epidemiol. 172 , 924 31. Braybrook, C., et al., 2001. The t box transcription factor gene tbx22 is mutated in x linked cleft palate and ankyloglossia. Nat Genet. 29 , 179 83.

PAGE 132

132 Bronfrenbrenner, U., 1979. The ecology of human development. Harvard Press, Cambridge. Brook, R. D., Rajagopalan, S., 2012. Chronic air pollution exposure and endothelial dysfunction: What you can't see -can harm you. J Am Coll Cardiol. 60 , 2167 9. Brunekreef, B., et al., 2009. Effects of long term exposure to traffic related air pollution on respiratory and cardiovascular mortality i n the netherlands: The nlcs air study. Res Rep Health Eff Inst. 5 71; discussion 73 89. Brunekreef, B., et al., 1997. Air pollution from truck traffic and lung function in children living near motorways. Epidemiology. 8 , 298 303. Cai, J., et al., 2015. D oes ambient co have protective effect for copd patient? Environ Res. 136 , 21 6. Canfield, M. A., et al., 2006a. National estimates and race/ethnic specific variation of selected birth defects in the united states, 1999 2001. Birth Defects Res A Clin Mol T eratol. 76 , 747 56. Canfield, M. A., et al., 2006b. Residential mobility patterns and exposure misclassification in epidemiologic studies of birth defects. J Expo Sci Environ Epidemiol. 16 , 538 43. Carlson, L., et al., 2013. Elevated infant mortality rat es among oral cleft and isolated oral cleft cases: a meta analysis of studies from 1943 to 2010. Cleft Palate Craniofac J. 50, 2 12. Carmichael, S. L., et al., 2009. Socioeconomic measures, orofacial clefts, and conotruncal heart defects in california. Bi rth Defects Res A Clin Mol Teratol. 85 , 850 7. Carmichael, S. L., et al., 2003. Socio economic status and risk of conotruncal heart defects and orofacial clefts. Paediatr Perinat Epidemiol. 17 , 264 71. Carmichael, S. L., et al., 2007. Maternal food insec urity is associated with increased risk of certain birth defects. J Nutr. 137 , 2087 92. Carrillo, A. E., et al., 2009. Effects of secondhand smoke on thyroid function. Inflamm Allergy Drug Targets. 8 , 359 63. Cassee, F. R., et al., 2013. Particulate matt er beyond mass: Recent health evidence on the role of fractions, chemical constituents and sources of emission. Inhal Toxicol. 25 , 802 12.

PAGE 133

133 Cassell, C. H., et al., 2008. Health care expenditures among medicaid enrolled children with and without orofacial clefts in north carolina, 1995 2002. Birth Defects Res A Clin Mol Teratol. 82 , 785 94. CDC, 1992. Recommendations for the use of folic acid to reduce the number of cases of spina bifida and other neural tube defects. MMWR Recomm Rep. 41 , 1 7. CDC, 1993. From the centers for disease control and prevention. Recommendations for use of folic acid to reduce number of spina bifida cases and other neural tube defects. JAMA. 269 , 1233, 1236 8. CDC, 2004. Spina bifida and anencephaly before and after folic acid m andate -united states, 1995 1996 and 1999 2000. MMWR Morb Mortal Wkly Rep. 53 , 362 5. CDC, 2010. How tobacco smoke causes disease: The biology and behavioral basis for smoking attributable disease: A report of the surgeon general. US Department of Human H ealth and Services, Atlanta (GA). CDC, Facts about cleft lip and cleft palate. Vol. 2014. Division of Birth Defects and Developmental Disabilities, NCBDDD, Centers for Disease Control and Prevention, Centers for Disease Control and Prevention, 2014. Chen , E. K., et al., 2014. Effects of air pollution on the risk of congenital anomalies: A systematic review and meta analysis. Int J Environ Res Public Health. 11 , 7642 68. Chevrier, C., et al., 2006. Occupational exposure to organic solvent mixtures during pregnancy and the risk of non syndromic oral clefts. Occup Environ Med. 63 , 617 23. Chin, B. Y., Otterbein, L. E., 2009. Carbon monoxide is a poison... To microbes! Co as a bactericidal molecule. Curr Opin Pharmacol. 9 , 490 500. Cho, J., et al., 2014. Ai r pollution as a risk factor for depressive episode in patients with cardiovascular disease, diabetes mellitus, or asthma. J Affect Disord. 157 , 45 51. Chum, A., O' Campo, P., 2013. Contextual determinants of cardiovascular diseases: Overcoming the reside ntial trap by accounting for non residential context and duration of exposure. Health Place. 24 , 73 9. Chung, K. C., et al., 2000. Maternal cigarette smoking during pregnancy and the risk of having a child with cleft lip/palate. Plast Reconstr Surg. 105 , 485 91. Clark, J. D., et al., 2003. Socioeconomic status and orofacial clefts in scotland, 1989 to 1998. Cleft Palate Craniofac J. 40 , 481 5.

PAGE 134

134 Colais, P., et al., 2012. Particulate air pollution and hospital admissions for cardiac diseases in potentially sensitive subgroups. Epidemiology. 23 , 473 81. Conrad, A. L., et al., 2014. Reading in subjects with an oral cleft: Speech, hearing and neuropsychological skills. Neuropsychology. 28 , 415 22. Cordier, S., et al., 2004. Risk of congenital anomalies in the vicinity of municipal solid waste incinerators. Occup Environ Med. 61 , 8 15. Cordier, S., et al., 2010. Maternal residence near municipal waste incinerators and the risk of urinary tract birth defects. Occup Environ Med. 67 , 493 9. Cornel, M. C., et al. , 1996. Population based birth defect and risk factor surveillance: Data from the northern netherlands. Int J Risk Saf Med. 8 , 197 209. Correa, A., et al., 2008. Diabetes mellitus and birth defects. Am J Obstet Gynecol. 199 , 237 e1 9. Croen, L. A., et al ., 1998. Racial and ethnic variations in the prevalence of orofacial clefts in california, 1983 1992. Am J Med Genet. 79 , 42 7. Dadvand, P., et al., 2011a. Ambient air pollution and congenital heart disease: A register based study. Environ Res. 111 , 435 4 1. Dadvand, P., et al., 2011b. Association between maternal exposure to ambient air pollution and congenital heart disease: A register based spatiotemporal analysis. Am J Epidemiol. 173 , 171 82. Damiano, P., et al., 2009. Demographic characteristics, car e, and outcomes for children with oral clefts in three states using participants from the national birth defects prevention study. Cleft Palate Craniofac J. 46 , 575 82. Das, S. K., et al., 1995. Epidemiology of cleft lip and cleft palate in mississippi. S outh Med J. 88 , 437 42. de Vries, I. A., et al., 2014. Prevalence of feeding disorders in children with cleft palate only: A retrospective study. Clin Oral Investig. 18 , 1507 15. Delfino, R. J., et al., 2005. Potential role of ultrafine particles in asso ciations between airborne particle mass and cardiovascular health. Environ Health Perspect. 113 , 934 46. Demetriou, C. A., et al., 2012. Biomarkers of ambient air pollution and lung cancer: A systematic review. Occup Environ Med. 69 , 619 27.

PAGE 135

135 Deodhar, J., Jarad, R., 1999. Study of the prevalence of and high risk factors for fetal malnutrition in term newborns. Ann Trop Paediatr. 19 , 273 7. DeRoo, L. A., et al., 2003. Orofacial cleft malformations: Associations with maternal and infant characteristics in w ashington state. Birth Defects Res A Clin Mol Teratol. 67 , 637 42. Diez Roux, A. V., 2001. Investigating neighborhood and area effects on health. Am J Public Health. 91 , 1783 9. Dolk, H., et al., 2010. Ambient air pollution and risk of congenital anomalies in england, 1991 1999. Occup Environ Med. 67 , 223 7. Dreier, J. W., et al., 2014. Systematic review and meta analyses: Fever in pregnancy and health impacts in the offspring. Pediatrics. 133 , e674 88. Dubowsky, S. D., et al., 2006. Diabetes, ob esity, and hypertension may enhance associations between air pollution and markers of systemic inflammation. Environ Health Perspect. 114 , 992 8. Dubrey, S. W., et al., 2015. Carbon monoxide poisoning: An ancient and frequent cause of accidental death. Br J Hosp Med (Lond). 76 , 159 62. Durante, W., et al., 2006. Role of carbon monoxide in cardiovascular function. J Cell Mol Med. 10 , 672 86. Durning, P., et al., 2007. The relationship between orofacial clefts and material deprivation in wales. Cleft Palat e Craniofac J. 44 , 203 7. Duval, S., Tweedie, R., 2000. Trim and fill: A simple funnel plot based method of testing and adjusting for publication bias in meta analysis. Biometrics. 56 , 455 63. Dvivedi, J., Dvivedi, S., 2012. A clinical and demographic pr ofile of the cleft lip and palate in sub himalayan india: A hospital based study. Indian J Plast Surg. 45 , 115 20. Egger, M., et al., 1997. Bias in meta analysis detected by a simple, graphical test. BMJ. 315 , 629 34. Engel, L. S., et al., 2000. Maternal occupation in agriculture and risk of limb defects in washington state, 1980 1993. Scand J Work Environ Health. 26 , 193 8. English, P. B., et al., 2003. Changes in the spatial pattern of low birth weight in a southern california county: The role of indiv idual and neighborhood level factors. Soc Sci Med. 56 , 2073 88.

PAGE 136

136 EPA, Integrated science assessment for carbon monoxide (final report). Vol. EPA/600/R 09/019F. US Environmental Protection Agency, Washington, DC., 2010. Farhi, A., et al., 2014. The possible association between exposure to air pollution and the risk for congenital malformations. Environ Res. 135 , 173 80. Fell, D. B., et al., 2004. Residential mobility during pregnancy. Paediatr Perinat Epidemiol. 18 , 408 14. Feng, C., et al., 2014. Association between polymorphism of tgfa taq i and cleft lip and/or palate: A meta analysis. BMC Oral Health. 14 , 88. Figueiredo, R. F., et al., 2015. The role of the folic acid to the prevention of orofacial cleft: An epidemiological study. Oral Dis. 21 , 240 7. Fone, D., et al., 2013. Common mental disorders, neighbourhood income inequality and income deprivation: Small area multilevel analysis. Br J Psychiatry. 202 , 286 93. Forrester, M. B., Merz, R. D., 2004. Descriptive epidemiology of oral clefts in a multiethnic population, hawaii, 1986 2000. Cleft Palate Craniofac J. 41 , 622 8. Forrester, M. B., Merz, R. D., 2007. Risk of selected birth defects with prenatal illicit drug use, hawaii, 1986 2002. J Toxicol Environ Health A. 70 , 7 18. Fry, R. C., et al., 2014. Air toxics and epigenetic effects: Ozone altered micrornas in the sputum of human subjects. Am J Physiol Lung Cell Mol Physiol. 306 , L1129 37. Furuta, C., et al., 2004. Estrogenic activities of nitrophenols in diesel exhaust particles. Biol Re prod. 70 , 1527 33. Gent, J. F., et al., 2003. Association of low level ozone and fine particles with respiratory symptoms in children with asthma. JAMA. 290 , 1859 67. Ghio, A. J., et al., 2012. Composition of air pollution particles and oxidative stress in cells, tissues, and living systems. J Toxicol Environ Health B Crit Rev. 15 , 1 21. Ghosh, R., et al., 2007. Does the effect of air pollution on pregnancy outcomes differ by gender? A systematic review. Environ Res. 105 , 400 8. Gilboa, S. M., et al., 2005. Relation between ambient air quality and selected birth defects, seven county study, texas, 1997 2000. Am J Epidemiol. 162 , 238 52. Gill, E. A., et al., 2011. Air pollution and cardiovascular disease in the multi ethnic study of atherosclerosis. Pro g Cardiovasc Dis. 53 , 353 60.

PAGE 137

137 Goh, Y. I., et al., 2006. Prenatal multivitamin supplementation and rates of congenital anomalies: A meta analysis. J Obstet Gynaecol Can. 28 , 680 9. Goldberg, M. S., et al., 2000. Identifying subgroups of the general popula tion that may be susceptible to short term increases in particulate air pollution: A time series study in montreal, quebec. Res Rep Health Eff Inst. 7 113; discussion 115 20. Gomez, S. L., et al., 2012. Patient, hospital, and neighborhood factors associat ed with treatment of early stage breast cancer among asian american women in california. Cancer Epidemiol Biomarkers Prev. 21 , 821 34. Grewal, J., et al., 2008. Maternal periconceptional smoking and alcohol consumption and risk for select congenital anoma lies. Birth Defects Res A Clin Mol Teratol. 82 , 519 26. Grosen, D., et al., 2011. Risk of oral clefts in twins. Epidemiology. 22 , 313 9. Guxens, M., et al., 2012. Prenatal exposure to residential air pollution and infant mental development: Modulation by antioxidants and detoxification factors. Environ Health Perspect. 120 , 144 9. Hackshaw, A., et al., 2011. Maternal smoking in pregnancy and birth defects: A systematic review based on 173 687 malformed cases and 11.7 million controls. Hum Reprod Update. 17 , 589 604. Hamano, T., et al., 2013. Neighbourhood environment and stroke: A follow up study in sweden. PLoS One. 8 , e56680. Hansen, C. A., et al., 2009. Ambient air pollution and birth defects in brisbane, australia. PLoS One. 4 , e5408. Hartley, D., et al., 1994. Urban and rural differences in health insurance and access to care. J Rural Health. 10 , 98 108. Hartwig, A., et al., 2002. Interference by toxic metal ions with DNA repair processes and cell cycle control: Molecular mechanisms. Environ Healt h Perspect. 110 Suppl 5 , 797 9. Hashmi, S. S., et al., 2005. Prevalence of nonsyndromic oral clefts in texas: 1995 1999. Am J Med Genet A. 134 , 368 72. Healthy People 2020, Maternal, infant, and child health: Objectives. Vol. 2014. US Department of Health and Human Services, Washington, DC, 2014. Heeren, G. A., et al., 2003. Agricultural chemical exposures and birth defects in the eastern cape province, south africa: A case control study. Environ Health. 2 , 11.

PAGE 138

138 Higgins, J. P., et al., 2003. Measuri ng inconsistency in meta analyses. BMJ. 327 , 557 60. Hoek, G., et al., 2013. Long term air pollution exposure and cardio respiratory mortality: A review. Environ Health. 12 , 43. Holmes, L. B., et al., 2008. Increased frequency of isolated cleft palate i n infants exposed to lamotrigine during pregnancy. Neurology. 70 , 2152 8. Holmes, L. B., Hernandez Diaz, S., 2012. Newer anticonvulsants: Lamotrigine, topiramate and gabapentin. Birth Defects Res A Clin Mol Teratol. 94 , 599 606. Honein, M. A., et al., 20 01. Impact of folic acid fortification of the us food supply on the occurrence of neural tube defects. JAMA. 285 , 2981 6. Honein, M. A., et al., 2007. Maternal smoking and environmental tobacco smoke exposure and the risk of orofacial clefts. Epidemiology . 18 , 226 33. Howland, R. E., et al., 2015. Reliability of reported maternal smoking: Comparing the birth certificate to maternal worksheets and prenatal and hospital medical records, new york city and vermont, 2009. Matern Child Health J [Epub ahead of print].. Hwang, B. F., Jaakkola, J. J., 2008. Ozone and other air pollutants and the risk of oral clefts. Environ Health Perspect. 116 , 1411 5. Ibald Mulli, A., et al., 2001. Effects of air pollution on blood pressure: A population based approach. Am J P ublic Health. 91 , 571 7. Izedonmwen, O. M., et al., 2015. What is the risk of having offspring with cleft lip/palate in pre maternal obese/overweight women when compared to pre maternal normal weight women? A systematic review and meta analysis. J Oral Ma xillofac Res. 6 , e1. Jafarabadi, M., 2007. Episodic air pollution is associated with increased DNA fragmentation in human sperm without other changes in semen quality. Hum Reprod. 22 , 3263; author reply 3264. Jagomagi, T., et al., 2010. Mthfr and msx1 contribute to the risk of nonsyndromic cleft lip/palate. Eur J Oral Sci. 118 , 213 20. James, W. H., 2000. Are oral clefts a consequence of maternal hormone imbalance? Evidence from the sex ratios of sibs of probands. Teratology. 62 , 342 5.

PAGE 139

139 James, W. H., 2004. Further evidence that mammalian sex ratios at birth are partially controlled by parental hormone levels around the time of conception. Hum Reprod. 19 , 1250 6. James, W. H., 2008a. Evidence that mammalian sex ratios at birth are partially controlled by parental hormone levels around the time of conception. J Endocrinol. 198 , 3 15. James, W. H., 2008b. Further support for the hypothesis that parental hormone levels around the time of conception are associated with human sex ratios at birth. J Biosoc S ci. 40 , 855 61. Janssen, B. G., et al., 2013. Placental DNA hypomethylation in association with particulate air pollution in early life. Part Fibre Toxicol. 10 , 22. Janssen, B. G., et al., 2012. Placental mitochondrial DNA content and particulate air pol lution during in utero life. Environ Health Perspect. 120 , 1346 52. Ji, H., Khurana Hershey, G. K., 2012. Genetic and epigenetic influence on the response to environmental particulate matter. J Allergy Clin Immunol. 129 , 33 41. Jia, Z. L., et al., 2011. Maternal malnutrition, environmental exposure during pregnancy and the risk of non syndromic orofacial clefts. Oral Dis. 17 , 584 9. Kahle, D., Wickham, H., Ggmap: Spatial visualization with google maps and openstreetmap. Vol. 2015. The Comprehensive R Arc hive Network, The Comprehensive R Archive Network, 2015. Kahle, J. J., et al., 2014. Interaction effects of temperature and ozone on lung function and markers of systemic inflammation, coagulation, and fibrinolysis: A crossover study of healthy young volu nteers. Environ Health Perspect. 133(4), 310 316.. Kan, H., et al., 2007. Traffic exposure and lung function in adults: The atherosclerosis risk in communities study. Thorax. 62 , 873 9. Kan, H., et al., 2008. Season, sex, age, and education as modifiers of the effects of outdoor air pollution on daily mortality in shanghai, china: The public health and air pollution in asia (papa) study. Environ Health Perspect. 116 , 1183 8. Kanaka Gantenbein, C., et al., 2003. Endocrine related causes and consequences o f intrauterine growth retardation. Ann N Y Acad Sci. 997 , 150 7. Kannan, S., et al., 2006. Exposures to airborne particulate matter and adverse perinatal outcomes: A biologically plausible mechanistic framework for exploring potential effect modification by nutrition. Environ Health Perspect. 114 , 1636 42.

PAGE 140

140 Kelly, F. J., Fussell, J. C., 2011. Air pollution and airway disease. Clin Exp Allergy. 41 , 1059 71. Khoury, M. J., et al., 1988. Residential mobility during pregnancy: Implications for environmental t eratogenesis. J Clin Epidemiol. 41 , 15 20. Klimova, N. G., et al., 2013. Does carbon monoxide inhibit proinflammatory cytokine production by fetal membranes? J Perinat Med. 41 , 683 90. Kohelet, D., et al., 1990. Reduced platelet counts in neonatal respiratory distress syndrome. Biol Neonate. 57 , 334 42. Kohli, S. S., Kohli, V. S., 2012. A comprehensive review of the genetic basis of cleft lip and palate. J Oral Maxillofac Pathol. 16 , 64 72. Kondo, S., et al., 2002. Mutations in irf6 cause van der woude and popliteal pterygium syndromes. Nat Genet. 32 , 285 9. Koukoura, O., et al., 2012. DNA methylation in the human placenta and fetal growth (review). Mol Med Rep. 5 , 883 9. Krieger, N., et al., 2003a. Race/ethnicity, gender, and monitoring socioeco nomic gradients in health: A comparison of area based socioeconomic measures -the public health disparities geocoding project. Am J Public Health. 93 , 1655 71. Krieger, N., et al., 2003b. Choosing area based socioeconomic measures to monitor social inequa lities in low birth weight and childhood lead poisoning: The public health disparities geocoding project (us). J Epidemiol Community Health. 57 , 186 99. Krishnan, R. M., et al., 2012. Vascular responses to long and short term exposure to fine particulate matter: Mesa air (multi ethnic study of atherosclerosis and air pollution). J Am Coll Cardiol. 60 , 2158 66. Kulldorff, M., 1997. A spatial scan statistic. Communications in Statistics Theory and methods. 26. Langlois, P. H., et al., 2009a. Maternal resi dential proximity to waste sites and industrial facilities and conotruncal heart defects in offspring. Paediatr Perinat Epidemiol. 23 , 321 31. Langlois, P. H., et al., 2013. Maternal occupational exposure to polycyclic aromatic hydrocarbons and risk of oral cleft affected pregnancies. Cleft Palate Craniofac J. 50 , 337 46.

PAGE 141

141 Langlois, P. H., et al., 2010. Occurrence of conotruncal heart birth defects in texas: A comparison of urban/rural classifications. J Rural Health. 26 , 164 74. Langlois, P. H., et al. , 2009b. Urban versus rural residence and occurrence of septal heart defects in texas. Birth Defects Res A Clin Mol Teratol. 85 , 764 72. Laumbach, R., et al., 2015. What can individuals do to reduce personal health risks from air pollution? J Thorac Dis. 7 , 96 107. Laumbach, R. J., Kipen, H. M., 2012. Respiratory health effects of air pollution: Update on biomass smoke and traffic pollution. J Allergy Clin Immunol. 129 , 3 11; quiz 12 3. Lebby, K. D., et al., 2010. Maternal factors and disparities associa ted with oral clefts. Ethn Dis. 20 , S1 146 9. Lee, P. C., et al., 2012. Ambient air pollution exposure and blood pressure changes during pregnancy. Environ Res. 117 , 46 53. Lee, P. C., et al., 2011. Particulate air pollution exposure and c reactive prote in during early pregnancy. Epidemiology. 22 , 524 31. Leite, I. C., et al., 2002. Chemical exposure during pregnancy and oral clefts in newborns. Cad Saude Publica. 18 , 17 31. Leite, M., et al., 2014. Maternal smoking in pregnancy and risk for congenital malformations: Results of a danish register based cohort study. Acta Obstet Gynecol Scand. 93 , 825 34. Levine, R. J., et al., 2004. Circulating angiogenic factors and the risk of preeclampsia. N Engl J Med. 350 , 672 83. Li, X., et al., 2013. Geographic a nd urban rural disparities in the total prevalence of neural tube defects and their subtypes during 2006 2008 in china: A study using the hospital based birth defects surveillance system. BMC Public Health. 13 , 161. Lie, R. T., et al., 2008. Maternal smok ing and oral clefts: The role of detoxification pathway genes. Epidemiology. 19 , 606 15. Lin, K. J., et al., 2012. Maternal exposure to amoxicillin and the risk of oral clefts. Epidemiology. 23 , 699 705. Little, J., et al., 2004. Tobacco smoking and oral clefts: A meta analysis. Bull World Health Organ. 82 , 213 8.

PAGE 142

142 Little, J., et al., 2008. Folate and clefts of the lip and palate -a u.K. based case control study: Part i: Dietary and supplemental folate. Cleft Palate Craniofac J. 45 , 420 7. Lodrup Carlsen , K. C., et al., 2006. Soluble cd14 at 2 yr of age: Gender related effects of tobacco smoke exposure, recurrent infections and atopic diseases. Pediatr Allergy Immunol. 17 , 304 12. Loomis, D., et al., 1999. Air pollution and infant mortality in mexico cit y. Epidemiology. 10 , 118 23. Lupo, P. J., et al., 2011. Maternal exposure to ambient levels of benzene and neural tube defects among offspring: Texas, 1999 2004. Environ Health Perspect. 119 , 397 402. Ma, J., et al., 2015. Parental health and social supp ort in the first trimester of pregnancy and the risk of oral clefts: A questionnaire based, case control study. Plast Reconstr Surg. 135 , 212 8. Marcusson, A., et al., 2001. Quality of life in adults with repaired complete cleft lip and palate. Cleft Pala te Craniofac J. 38 , 379 85. Marshall, E. G., et al., 2010. Oral cleft defects and maternal exposure to ambient air pollutants in new jersey. Birth Defects Res A Clin Mol Teratol. 88 , 205 15. Martin, J. A., et al., 2008. Annual summary of vital statistics : 2006. Pediatrics. 121 , 788 801. Mateja, W. A., et al., 2012. The association between maternal alcohol use and smoking in early pregnancy and congenital cardiac defects. J Womens Health (Larchmt). 21 , 26 34. May, J. M., 1958. Ecology of disease in world health. U S Armed Forces Med J. 9 , 781 94. McKinney, C. M., et al., 2013. Micronutrients and oral clefts: A case control study. J Dent Res. 92 , 1089 94. McMillan, N., Holland, D. M., Morara, M., and Feng, J., 2010. Combining numerical model output and p articulate data using bayesian space time modeling. Environmetrics. 21 , 48 65. Medina Ramon, M., Schwartz, J., 2008. Who is more vulnerable to die from ozone air pollution? Epidemiology. 19 , 672 9.

PAGE 143

143 Messer, L. C., et al., 2010. Urban rural residence and the occurrence of cleft lip and cleft palate in texas, 1999 2003. Ann Epidemiol. 20 , 32 9. Messer, L. C., et al., 2008. Socioeconomic domains and associations with preterm birth. Soc Sci Med. 67 , 1247 57. Metcalfe, A., et al., 2011. The association betwe en neighbourhoods and adverse birth outcomes: A systematic review and meta analysis of multi level studies. Paediatr Perinat Epidemiol. 25 , 236 45. Meyer, K. A., et al., 2003. Low maternal alcohol consumption during pregnancy and oral clefts in offspring: The slone birth defects study. Birth Defects Res A Clin Mol Teratol. 67 , 509 14. Mines, D., et al., 2014. Topiramate use in pregnancy and the birth prevalence of oral clefts. Pharmacoepidemiol Drug Saf. 23 , 1017 25. Miousse, I. R., et al., 2014. Epigene tic alterations induced by ambient particulate matter in mouse macrophages. Environ Mol Mutagen. Miraglia, S. G., et al., 2013. Follow up of the air pollution and the human male to female ratio analysis in sao paulo, brazil: A times series study. BMJ Open . 3. Mittal, M., et al., 2013. Reactive oxygen species in inflammation and tissue injury. Antioxid Redox Signal . 20(7) , 1126 67 . Mobasher, Z., et al., 2013. Associations between ambient air pollution and hypertensive disorders of pregnancy. Environ Res. 123 , 9 16. Molina Solana, R., et al., 2013. Current concepts on the effect of environmental factors on cleft lip and palate. Int J Oral Maxillofac Surg. 42 , 177 84. Moraleda Cibrian, M., et al., 2014. Symptoms of sleep disordered breathing in children wi th craniofacial malformations. J Clin Sleep Med. 10 , 307 12. Mossey, P. A., et al., 2009. Cleft lip and palate. Lancet. 374 , 1773 85. Mossey, P. A., Modell, B., 2012. Epidemiology of oral clefts 2012: An international perspective. Front Oral Biol. 16 , 1 18. Murphy, S. R., et al., 2014. Ozone induced airway epithelial cell death, the neurokinin 1 receptor pathway, and the postnatal developing lung. Am J Physiol Lung Cell Mol Physiol. 307 , L471 81. Nagiah, S., et al., 2014. Oxidative stress and air pollution exposure during pregnancy: A molecular assessment. Hum Exp Toxicol. [Epub ahead of print]

PAGE 144

144 National Research Council, Research priorities for airborne particulate matter. Iv. Continuing research progress. . National Research Council, Washington, DC., 2004. Nieuwenhuijsen, M. J., et al., 2013. Environmental risk factors of pregnancy outcomes: A summary of recent meta analyses of epidemiological studies. Environ Health. 12 , 6. Ochoa Acuna, H., Carbajo, C., 2009. Risk of limb birth defects and moth er's home proximity to cornfields. Sci Total Environ. 407 , 4447 51. Olgun, N. S., et al., 2014. Carbon monoxide attenuates bacteria induced endothelin 1 expression in second trimester placental explants. Placenta. 35 , 351 8. Orbach, H., et al., 2013. Hyp ertension and antihypertensive drugs in pregnancy and perinatal outcomes. Am J Obstet Gynecol. 208 , 301 e1 6. Otterbein, L. E., et al., 2000. Carbon monoxide has anti inflammatory effects involving the mitogen activated protein kinase pathway. Nat Med. 6 , 422 8. Otterbein, L. E., et al., 1999. Carbon monoxide provides protection against hyperoxic lung injury. Am J Physiol. 276 , L688 94. Otterbein, L. E., et al., 2005. Carbon monoxide increases macrophage bacterial clearance through toll like receptor (tl r)4 expression. Cell Mol Biol (Noisy le grand). 51 , 433 40. Padula, A. M., et al., 2013a. The association of ambient air pollution and traffic exposures with selected congenital anomalies in the san joaquin valley of california. Am J Epidemiol. 177 , 1074 85. Padula, A. M., et al., 2013b. Ambient air pollution and traffic exposures and congenital heart defects in the san joaquin valley of california. Paediatr Perinat Epidemiol. 27 , 329 39. Pae, H. O., et al., 2004. Carbon monoxide produced by heme oxygena se 1 suppresses t cell proliferation via inhibition of il 2 production. J Immunol. 172 , 4744 51. Pardi, G., et al., 2002. Placental fetal interrelationship in iugr fetuses -a review. Placenta. 23 Suppl A , S136 41. Parker, S. E., et al., 2010. Updated national birth prevalence estimates for selected birth defects in the united states, 2004 2006. Birth Defects Res A Clin Mol Teratol. 88 , 1008 16.

PAGE 145

145 Pekkanen, J., et al., 2000. Daily concentrations of air pollution and plasma fibrinogen in london. Occup Environ Med. 57 , 818 22. Pereira, G., et al., 2012. Locally derived traffic related air pollution and fetal growth restriction: A retrospective cohort study. Occup Environ Med. 69 , 815 22. Pershagen, G., et al., 1995. Air pollution involving nitrogen dio xide exposure and wheezing bronchitis in children. Int J Epidemiol. 24 , 1147 53. Peters, A., et al., 1997. Increased plasma viscosity during an air pollution episode: A link to mortality? Lancet. 349 , 1582 7. Poletta, F. A., et al., 2007. Regional analys is on the occurrence of oral clefts in south america. Am J Med Genet A. 143A , 3216 27. Pope, C. A., 3rd, Dockery, D. W., 2006. Health effects of fine particulate air pollution: Lines that connect. J Air Waste Manag Assoc. 56 , 709 42. Proietti, E., et al. , 2013. Air pollution during pregnancy and neonatal outcome: A review. J Aerosol Med Pulm Drug Deliv. 26 , 9 23. Queisser Luft, A., et al., 2011. Birth defects in the vicinity of nuclear power plants in germany. Radiat Environ Biophys. 50 , 313 23. Raascho u Nielsen, O., Reynolds, P., 2006. Air pollution and childhood cancer: A review of the epidemiological literature. Int J Cancer. 118 , 2920 9. Rankin, J., et al., 2009. Maternal exposure to ambient air pollutants and risk of congenital anomalies. Environ R es. 109 , 181 7. Reefhuis, J., Honein, M. A., 2004. Maternal age and non chromosomal birth defects, atlanta -1968 2000: Teenager or thirty something, who is at risk? Birth Defects Res A Clin Mol Teratol. 70 , 572 9. Rios, R., et al., 1993. Susceptibility t o environmental pollutants among minorities. Toxicol Ind Health. 9 , 797 820. Risom, L., et al., 2005. Oxidative stress induced DNA damage by particulate air pollution. Mutat Res. 592 , 119 37. Ritz, B., et al., 2006. Air pollution and infant death in southern california, 1989 2000. Pediatrics. 118 , 493 502. Ritz, B., et al., 2002. Ambient air pollution and risk of birth defects in southern california. Am J Epidemiol. 155 , 17 25.

PAGE 146

146 Rodrigues, K., et al., 2009. Prevalence of orofacial clefts and social f actors in brazil. Braz Oral Res. 23 , 38 42. Romao, R., et al., 2013. The relationship between low birth weight and exposure to inhalable particulate matter. Cad Saude Publica. 29 , 1101 8. Root, E. D., et al., 2009. Evidence of localized clustering of gas troschisis births in north carolina, 1999 2004. Soc Sci Med. 68 , 1361 7. Rossner, P., Jr., et al., 2011. Genetic, biochemical, and environmental factors associated with pregnancy outcomes in newborns from the czech republic. Environ Health Perspect. 119 , 265 71. Rozendaal, A. M., et al., 2013. Periconceptional folic acid associated with an increased risk of oral clefts relative to non folate related malformations in the northern netherlands: A population based case control study. Eur J Epidemiol. 28 , 875 87. Rubes, J., et al., 2005. Episodic air pollution is associated with increased DNA fragmentation in human sperm without other changes in semen quality. Hum Reprod. 20 , 2776 83. Ryter, S. W., et al., 2006. Heme oxygenase 1/carbon monoxide: From basic sc ience to therapeutic applications. Physiol Rev. 86 , 583 650. Sarkar, D., 2012. Recurrent pregnancy loss in patients with thyroid dysfunction. Indian J Endocrinol Metab. 16 , S350 1. Sarrafzadegan, N., et al., 2012. The influence of gender and place of residence on cardiovascular diseases and their risk factors. The isfahan cohort study. Saudi Med J. 33 , 533 40. Schembari, A., et al., 2014. Traffic related air pollution and congenital anomalies in barcelona. Environ Health Perspect. 122 , 317 23. Schwartz, J., 1994. Air pollution and daily mortality: A review and meta analysis. Environ Res. 64 , 36 52. Selevan, S. G., et al., 2000. Semen quality and reproductive health of young czech men exposed to seasonal air pollution. Environ Health Perspect. 1 08 , 887 94. Semenza, J. C., et al., 2008. Public perception and behavior change in relationship to hot weather and air pollution. Environ Res. 107 , 401 11. Sexton, K., 1997. Sociodemographic aspects of human susceptibility to toxic chemicals: Do class an d race matter for realistic risk assessment? Environ Toxicol Pharmacol. 4 , 261 9.

PAGE 147

147 Sexton, K., et al., 1993. Air pollution health risks: Do class and race matter? Toxicol Ind Health. 9 , 843 78. Shahrukh Hashmi, S., et al., 2010. Maternal fever during earl y pregnancy and the risk of oral clefts. Birth Defects Res A Clin Mol Teratol. 88 , 186 94. Shang, Y., et al., 2013. Systematic review of chinese studies of short term exposure to air pollution and daily mortality. Environ Int. 54 , 100 11. Shariff Marco, S., et al., 2014. Impact of neighborhood and individual socioeconomic status on survival after breast cancer varies by race/ethnicity: The neighborhood and breast cancer study. Cancer Epidemiol Biomarkers Prev. 23(5) , 793 811 . Sharkhuu, T., et al., 2011. E ffect of maternal exposure to ozone on reproductive outcome and immune, inflammatory, and allergic responses in the offspring. J Immunotoxicol. 8 , 183 94. Sharma, R. K., Nanda, V., 2009. Problems of middle ear and hearing in cleft children. Indian J Plast Surg. 42 Suppl , S144 8. Sharp, A. N., et al., 2010. Placental apoptosis in health and disease. Am J Reprod Immunol. 64 , 159 69. Shaw, G. M., et al., 2006. Maternal nutrient intakes and risk of orofacial clefts. Epidemiology. 17 , 285 91. Shaw, G. M., et al., 1991. Isolated oral cleft malformations: Associations with maternal and infant characteristics in a california population. Teratology. 43 , 225 8. Shkoukani, M. A., et al., 2013. Cleft lip a comprehensive review. Front Pediatr. 1 , 53. Sienra Monge, J. J., et al., 2004. Antioxidant supplementation and nasal inflammatory responses among young asthmatics exposed to high levels of ozone. Clin Exp Immunol. 138 , 317 22. Simmons, C. J., et al., 2004. Birth defects in arkansas: Is folic acid f ortification making a difference? Birth Defects Res A Clin Mol Teratol. 70 , 559 64. Sioutas, C., et al., 2005. Exposure assessment for atmospheric ultrafine particles (ufps) and implications in epidemiologic research. Environ Health Perspect. 113 , 947 55. Slama, R., et al., 2008. Meeting report: Atmospheric pollution and human reproduction. Environ Health Perspect. 116 , 791 8.

PAGE 148

148 Slifkin, R. T., et al., 2004. The changing metropolitan designation process and rural america. J Rural Health. 20 , 1 6. Somers, C. M., Cooper, D. N., 2009. Air pollution and mutations in the germline: Are humans at risk? Hum Genet. 125 , 119 30. Somers, C. M., et al., 2004. Reduction of particulate air pollution lowers the risk of heritable mutations in mice. Science. 304 , 1008 10. Somers, C. M., et al., 2002. Air pollution induces heritable DNA mutations. Proc Natl Acad Sci U S A. 99 , 15904 7. Souza, L. T., et al., 2013. Msx1 gene and nonsyndromic oral clefts in a southern brazilian population. Braz J Med Biol Res. 46 , 555 8. So zen, M. A., et al., 2001. Mutation of pvrl1 is associated with sporadic, non syndromic cleft lip/palate in northern venezuela. Nat Genet. 29 , 141 2. Strickland, M. J., et al., 2009. Ambient air pollution and cardiovascular malformations in atlanta, georgi a, 1986 2003. Am J Epidemiol. 169 , 1004 14. Su, H. J., et al., 2013. Effects of vitamin c and e intake on peak expiratory flow rate of asthmatic children exposed to atmospheric particulate matter. Arch Environ Occup Health. 68 , 80 6. Su, P. Y., et al., 2 011a. Maternal thyroid function in the first twenty weeks of pregnancy and subsequent fetal and infant development: A prospective population based cohort study in china. J Clin Endocrinol Metab. 96 , 3234 41. Su, T. C., et al., 2011b. Progress of ambient air pollution and cardiovascular disease research in asia. Prog Cardiovasc Dis. 53 , 369 78. Suzuki, K., et al., 2000. Mutations of pvrl1, encoding a cell cell adhesion molecule/herpesvirus receptor, in cleft lip/palate ectodermal dysplasia. Nat Genet. 25 , 427 30. Takeda, K., et al., 2004. Endocrine disrupting activity of chemicals in diesel exhaust and diesel exhaust particles. Environ Sci. 11 , 33 45. Tannure, P. N., et al., 2012. Prevalence of dental anomalies in nonsyndromic individuals with cleft lip and palate: A systematic review and meta analysis. Cleft Palate Craniofac J. 49 , 194 200. Tao, F., et al., 2003. Reactive oxygen species in pulmonary inflammation by ambient particulates. Free Radic Biol Med. 35 , 327 40.

PAGE 149

149 Tomei, G., et al., 2007. Plasma 1 7 alpha oh progesterone in male workers exposed to traffic pollutants. Ind Health. 45 , 170 6. Tong, L., et al., 2015. The association between air pollutants and morbidity for diabetes and liver diseases modified by sexes, ages, and seasons in tianjin, chi na. Environ Sci Pollut Res Int. 22 , 1215 9. Tran, D. Q., et al., 1996. The anti estrogenic activity of selected polynuclear aromatic hydrocarbons in yeast expressing human estrogen receptor. Biochem Biophys Res Commun. 229 , 101 8. Truong, K. D., Ma, S., 2006. A systematic review of relations between neighborhoods and mental health. J Ment Health Policy Econ. 9 , 137 54. Turgut, S., et al., 2005. Effects of cadmium and zinc on plasma levels of growth hormone, insulin like growth factor i, and insulin like growth factor binding protein 3. Biol Trace Elem Res. 108 , 197 204. Turner, M. C., et al., 2014. Interactions between cigarette smoking and fine particulate matter in the risk of lung cancer mortality in cancer prevention study ii. Am J Epidemiol. 180 , 11 45 9. Uzoigwe, J. C., et al., 2013. The emerging role of outdoor and indoor air pollution in cardiovascular disease. N Am J Med Sci. 5 , 445 453. van den Boogaard, E., et al., 2011. Significance of (sub)clinical thyroid dysfunction and thyroid autoimmunit y before conception and in early pregnancy: A systematic review. Hum Reprod Update. 17 , 605 19. van Vliet, P., et al., 1997. Motor vehicle exhaust and chronic respiratory symptoms in children living near freeways. Environ Res. 74 , 122 32. Vieira, A. R., et al., 2002. Maternal age and oral clefts: A reappraisal. Oral Surg Oral Med Oral Pathol Oral Radiol Endod. 94 , 530 5. Vinikoor Imler, L. C., et al., 2013. Early prenatal exposure to air pollution and its associations with birth defects in a state wide b irth cohort from north carolina. Birth Defects Res A Clin Mol Teratol. 97 , 696 701. Vinikoor Imler, L. C., et al., 2015. An exploratory analysis of the relationship between ambient ozone and particulate matter concentrations during early pregnancy and sel ected birth defects in texas. Environ Pollut. 202 , 1 6. Vrijheid, M., et al., 2002. Hazard potential ranking of hazardous waste landfill sites and risk of congenital anomalies. Occup Environ Med. 59 , 768 76.

PAGE 150

150 Vrijheid, M., et al., 2000. Socioeconomic ineq ualities in risk of congenital anomaly. Arch Dis Child. 82 , 349 52. Vrijheid, M., et al., 2011. Ambient air pollution and risk of congenital anomalies: A systematic review and meta analysis. Environ Health Perspect. 119 , 598 606. Wang, Y., et al., 2006. Problems in using birth certificate files in the capture recapture model to estimate the completeness of case ascertainment in a population based birth defects registry in new york state. Birth Defects Res A Clin Mol Teratol. 76 , 772 7. Wasserman, C. R., et al., 1998. Socioeconomic status, neighborhood social conditions, and neural tube defects. Am J Public Health. 88 , 1674 80. Wehby, G. L., Cassell, C. H., 2010. The impact of orofacial clefts on quality of life and healthcare use and costs. Oral Dis. 16 , 3 10. Wehby, G. L., et al., 2014a. Academic achievement of children and adolescents with oral clefts. Pediatrics. 133 , 785 92. Wehby, G. L., et al., 2014b. Academic achievement of children and adolescents with oral clefts. Pediatrics. Wehby, G. L., Mur ray, J. C., 2010. Folic acid and orofacial clefts: A review of the evidence. Oral Dis. 16 , 11 9. Wehby, G. L., et al., 2012. The effects of oral clefts on hospital use throughout the lifespan. BMC Health Serv Res. 12 , 58. Weiss, J., et al., 2009. Hospital use and associated costs of children aged zero to two years with craniofacial malformations in massachusetts. Birth Defects Res A Clin Mol Teratol. 85 , 925 34. Wen, X. J., et al., 2009. Association between media alerts of air quality index and ch ange of outdoor activity among adult asthma in six states, brfss, 2005. J Community Health. 34 , 40 6. Whyatt, R. M., et al., 1998. Relationship between ambient air pollution and DNA damage in polish mothers and newborns. Environ Health Perspect. 106 Suppl 3 , 821 6. Williams, L. J., et al., 2005. Decline in the prevalence of spina bifida and anencephaly by race/ethnicity: 1995 2002. Pediatrics. 116 , 580 6. Wilson, J. L., et al., 2012. Antibacterial effects of carbon monoxide. Curr Pharm Biotechnol. 13 , 76 0 8.

PAGE 151

151 Woodruff, T. J., et al., 1997. The relationship between selected causes of postneonatal infant mortality and particulate air pollution in the united states. Environ Health Perspect. 105 , 608 12. Wu, S., et al., 2012. Chemical constituents of ambient particulate air pollution and biomarkers of inflammation, coagulation and homocysteine in healthy adults: A prospective panel study. Part Fibre Toxicol. 9 , 49. Wyszynski, D. F., et al., 1997. Maternal cigarette smoking and oral clefts: A meta analysis. C left Palate Craniofac J. 34 , 206 10. Xu, X., et al., 2013. Association between ozone exposure and onset of stroke in allegheny county, pennsylvania, USA, 1994 2000. Neuroepidemiology. 41 , 2 6. Yamamoto, S. S., et al., 2014. A systematic review of air pollution as a risk factor for cardiovascular disease in south asia: Limited evidence from india and pakistan. Int J Hyg Environ Health. 217 , 133 44. Yan, M., et al., 2013. Meta analysis of the chinese studies of the association between ambient ozone and mortality. Chemosphere. 93 , 899 905. Yauk, C., et al., 2008. Germ line mutations, DNA damage, and global hypermethylation in mice exposed to particulate air pollution in an urban/industrial location. Proc Natl Acad Sci U S A. 105 , 605 10. Yazdy, M. M., e t al., 2008. Use of special education services by children with orofacial clefts. Birth Defects Res A Clin Mol Teratol. 82 , 147 54. Yazdy, M. M., et al., 2015. Spatial analysis of gastroschisis in massachusetts and texas. Ann Epidemiol. 25 , 7 14. Yen, I. H., et al., 2009. Neighborhood environment in studies of health of older adults: A systematic review. Am J Prev Med. 37 , 455 63. Yu, D., et al., 2014. Maternal socioeconomic status and the risk of congenital heart defects in offspring: A meta analysis of 33 studies. PLoS One. 9 , e111056. Zanobetti, A., Schwartz, J., 2000. Race, gender, and social status as modifiers of the effects of pm10 on mortality. J Occup Environ Med. 42 , 469 74. Zanobetti, A., et al., 2000. Are there sensitive subgroups for the ef fects of airborne particles? Environ Health Perspect. 108 , 841 5. Zeka, A., et al., 2006. Individual level modifiers of the effects of particulate matter on daily mortality. Am J Epidemiol. 163 , 849 59.

PAGE 152

152 Zhang, B., et al., 2011. Maternal cigarette smoking and the associated risk of having a child with orofacial clefts in china: A case control study. J Craniomaxillofac Surg. 39 , 313 8.

PAGE 153

153 BIOGRAPHICAL SKETCH Sandie Uyen Ha received her Doctor of Philosophy from t he Department of Epidemiology at the Uni versity of Florida in August 2015. She was a teaching and research fellow under the University of Florida Graduate School Fellowshi p. She also received her Master of Public Health (MPH) with a concentration in ep idemiology from the University of Florida in 2011, and Bachelor of Science in b iology from the University of Washington in 2008. After the completion of her MPH, Sandie became an intern at the University o f Florida Family Data Center, where she evaluated the Florida Family Planning Waiver on its eff ectiveness in preventing adverse birth outcomes. Her findings were directly used by the State to pursue the renewal of the Waiver for the next cycle. Among her broad interest in maternal and child health, Sandie is particularly interested in studying envir onmental determinant s of perinatal health outcomes. She will be joining the National Institute of Child Health and Development (NICHD), where she will be continuing her research.