Birth Outcomes of Immigrants to Urban Ontario. A ... · protocols since my Masters programs, providing useful tips to improve my theses, and my Departmental reviewers Dionne Gesink

Birth Outcomes of Immigrants to Urban Ontario. A population-based study.

by

Marcelo Luis Urquia

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy

Dalla Lana School of Public Health Sciences

University of Toronto

© Copyright by Marcelo Luis Urquia (2009)

ii

Birth Outcomes of Immigrants to Urban Ontario.

A population-based study.

Marcelo Luis Urquia

Doctor of Philosophy (Epidemiology)

Dalla Lana School of Public Health Sciences University of Toronto

2009

Abstract

The total number of births among immigrants is on the rise and currently exceeds one fifth of

live births within industrialized countries. The relation between adverse birth outcomes and

migration remains unclear.

The objectives of this thesis are to undertake a literature review to clarify the relation between

migration and adverse birth outcomes, and to examine the interplay between duration of

residence, maternal country of origin, and the residential environment using data on immigrants

to Ontario Census Metropolitan Areas. The findings indicate that:

a) Analyzing disparities in birth outcomes by migrant status with migrants defined as a single

category is not informative. Rather, ethnicity and country of origin are important predictors of

birth outcomes among immigrants.

b) Duration of residence is linearly associated with low infant birth weight and preterm birth,

mainly driven by decreases in gestational age with prolonged stay in Canada.

iii

c) The detrimental effects of long duration of residence on preterm birth are modestly attenuated,

but not prevented, among immigrants living in urban neighbourhoods characterized by low

material deprivation.

d) Neighbourhood material deprivation has little, if any, influence on birth outcomes of recent

immigrants, and only becomes influential after 15 years of stay in Canada. Maternal world

region of origin constitutes a stronger predictor of adverse birth outcomes among recent

immigrants.

These findings stress the importance of the maternal country of birth and duration of residence as

key predictors of immigrants’ health. They also support further research aimed at clarifying the

nature of the association between time spent in Canada after migration and decreases in

gestational age at delivery, and the identification of immigrant groups at high risk of adverse

birth outcomes, based on these two key predictors.

iv

Acknowledgments

This thesis is the final product of a process that started some years ago, when I was doing my

Masters program. Most of the people directly or indirectly supporting my work remained the

same along the way. In first place, I thank my supervisor John Frank for his continuous financial,

academic, and social support. John encouraged me to go deeper in my scientific inquiry by

stressing my achievements rather than my limitations. He always had time to assist me when

needed, despite his extremely busy schedule. My other committee members, Rick Glazier and

Rahim Moineddin, were also readily available every time I needed them. My co-supervisor, Rick

Glazier, was the key person to identify and access data sources and research opportunities. He

provided me with student space and administrative support at the St. Michael Hospital and

supported my data analyses at ICES. Now I thank both John and Rick for having encouraged me

to pursue this thesis instead of other less fruitful topics I was attracted to. Rahim not only served

as my biostatistician but also as an insightful interlocutor to discuss the relation between

migration and health. I learnt more than biostatistics from him. I hope my committee had enjoyed

our meetings as much as I did. I also would like to acknowledge the ROAM Collaboration,

particularly Anita Gagnon. My involvement with this group of researchers certainly promoted

my interest in the area of research of this thesis. Alex Kopp and Kinwah Fung from ICES also

deserve my gratitude for being so helpful to clarify my concerns about the administrative data

sources. Discussions with other people, such as Flora Matheson and all the participants of the

Time Trends meetings, also fuelled my enthusiasm in trying a little bit harder. I also have to

acknowledge Marisa Creatore for sharing the long process of understanding the immigrant data. I

am also grateful to Doug Manuel and Patricia O’Campo, who have kindly accepted to review my

protocols since my Masters programs, providing useful tips to improve my theses, and my

Departmental reviewers Dionne Gesink Law and Joel Ray, and my external reviewer K.S.

Joseph, who also provided helpful comments and corrections. I would also like to thank staff at

the Institute of Population and Public Health (IPPH), Institute for Work and Health (IWH), St.

Michael Hospital, and ICES, which routine work indirectly supported the progress of this thesis:

Gail Bryant and Vera Ndaba at IPPH, Sandra Sinclair and Mary Cicinelli at the IWH, J.R.,

Jackson Wong, and Donna Hoppenheim at ICES, and Claudeth White at the St. Michael

Hospital. Finally I thank my wife Denise for her intangible, but not negligible, contribution to

this thesis through her positive influence on my self.

v

Table of Contents TABLE OF CONTENTS............................................................................................................................................ V

LIST OF TABLES ..................................................................................................................................................... IX

LIST OF FIGURES ...................................................................................................................................................XI

LIST OF APPENDICES.......................................................................................................................................... XII

ABBREVIATIONS .................................................................................................................................................XIII

CHAPTER 1 INTRODUCTION ................................................................................................................................ 1

1.1. BACKGROUND ................................................................................................................................................... 1

1.2. STUDY OBJECTIVES......................................................................................................................................... 4

1.3. INVOLVEMENT/PARTICIPATION OF AUTHOR AND CO-AUTHORS IN THE RESEARCH............ 5

1.4. ORGANIZATION OF THE THESIS ................................................................................................................. 6

1.4.1. RATIONALE FOR THE OBJECTIVES ................................................................................................................. 6 1.4.2. OVERVIEW OF THE RESEARCH ....................................................................................................................... 9

1.5. REFERENCES.................................................................................................................................................... 12

CHAPTER 2 INTERNATIONAL MIGRATION AND ADVERSE BIRTH OUTCOMES: ROLE OF

ETHNICITY, REGION OF ORIGIN AND DESTINATION ............................................................................... 15

ABSTRACT ................................................................................................................................................................ 15

2.1. INTRODUCTION............................................................................................................................................... 16

2.2. METHODS .......................................................................................................................................................... 17

2.2.1. STUDY POPULATION...................................................................................................................................... 17 2.2.2. SEARCH AND STUDY SELECTION CRITERIA.................................................................................................. 17 2.2.3. DATA EXTRACTION ....................................................................................................................................... 19 2.2.4. STATISTICAL ANALYSES ............................................................................................................................... 19

2.3. RESULTS............................................................................................................................................................. 21

2.3.1. MIGRATION AND RACE/ETHNICITY .............................................................................................................. 22 2.3.2. MIGRATION AND WORLD REGIONS .............................................................................................................. 25

2.4. DISCUSSION ...................................................................................................................................................... 27

vi

2.4.1. MIGRATION AND ETHNIC DISPARITIES ........................................................................................................ 31 2.4.2. MIGRATION AND REGION OF ORIGIN AND DESTINATION ............................................................................ 32 2.4.3. FURTHER RESEARCH .................................................................................................................................... 33

2.5. REFERENCES.................................................................................................................................................... 35

CHAPTER 3 INCREASE IN PRETERM BIRTHS ASSOCIATED WITH DURATION OF RESIDENCE

AMONG IMMIGRANTS LIVING IN ONTARIO METROPOLITAN AREAS................................................ 46

ABSTRACT ................................................................................................................................................................ 46

3.1. INTRODUCTION............................................................................................................................................... 47

3.2. METHODS .......................................................................................................................................................... 49

3.2.1. DATA SOURCES.............................................................................................................................................. 49 3.2.2. OUTCOMES .................................................................................................................................................... 51 3.2.3. PREDICTORS.................................................................................................................................................. 51 3.2.4. STATISTICAL ANALYSES ............................................................................................................................... 53

3.3. RESULTS............................................................................................................................................................. 54

3.3.1. SENSITIVITY ANALYSES ................................................................................................................................ 61

3.4. DISCUSSION ...................................................................................................................................................... 63

3.5. REFERENCES.................................................................................................................................................... 70

CHAPTER 4 THE INTERPLAY BETWEEN IMMIGRANTS’ COUNTRY OF BIRTH AND

NEIGHBOURHOOD DEPRIVATION ON BIRTH OUTCOMES ...................................................................... 78

ABSTRACT ................................................................................................................................................................ 78

4.1. INTRODUCTION............................................................................................................................................... 79

4.2. METHODS .......................................................................................................................................................... 80

4.2.1. DATA.............................................................................................................................................................. 80 4.2.2. OUTCOMES .................................................................................................................................................... 81 4.2.3. PREDICTORS.................................................................................................................................................. 82 4.2.4. STATISTICAL ANALYSES ............................................................................................................................... 83 4.2.5. MODELING STRATEGY .................................................................................................................................. 84

4.3. RESULTS............................................................................................................................................................. 86

4.4. DISCUSSION ...................................................................................................................................................... 92

vii

4.5. REFERENCES.................................................................................................................................................... 98

CHAPTER 5 THE DIFFERENTIAL DETERIORATION OF PRETERM BIRTH AMONG URBAN

IMMIGRANTS BY NEIGHBOURHOOD DEPRIVATION .............................................................................. 107

ABSTRACT .............................................................................................................................................................. 107

5.1. INTRODUCTION............................................................................................................................................. 108

5.2. METHODS ........................................................................................................................................................ 109

5.3. RESULTS........................................................................................................................................................... 110

5.4. DISCUSSION .................................................................................................................................................... 112

5.5. REFERENCES.................................................................................................................................................. 113

CHAPTER 6 DISCUSSION.................................................................................................................................... 116

6.1. MAIN FINDINGS ............................................................................................................................................. 116

6.3. IMPLICATIONS FOR PRACTICE AND FUTURE RESEARCH............................................................. 121

6.4. UNANSWERED QUESTIONS AND FUTURE RESEARCH ..................................................................... 124

6.4.1. WHY IS DURATION OF RESIDENCE ASSOCIATED WITH PRETERM BIRTH?..................................................... 124 6.4.2. IS THE ASSOCIATION BETWEEN TIME SINCE MIGRATION AND PRETERM BIRTH MERELY A CANADIAN

PHENOMENON? ...................................................................................................................................................... 129 6.4.3. IS TIME SINCE MIGRATION ASSOCIATED WITH OTHER PREGNANCY-RELATED OUTCOMES? ........................ 130 6.4.4. WHY DO SOME MIGRANT GROUPS EXPERIENCE POOR OUTCOMES AND OTHERS DO NOT? ........................... 131

6.5. CONCLUDING REMARK.............................................................................................................................. 132

APPENDICES .......................................................................................................................................................... 140

APPENDIX 2.A. SEARCH STRATEGY............................................................................................................... 140

APPENDIX 3.A. DATA SOURCES ....................................................................................................................... 143

APPENDIX 3.B. MEASUREMENT OF STILLBIRTHS AND MULTIPLE BIRTHS USING THE

DISCHARGE ABSTRACT DATABASE .............................................................................................................. 151

APPENDIX 3.C. FLOWCHART DATA EXCLUSIONS .................................................................................... 162

APPENDIX 3.D. COVARIATE ADJUSTMENT BASED ON DIRECTED ACYCLIC GRAPHS (DAGS).. 164

viii

APPENDIX 3.E. USING A COHORT APPROACH TO RULE OUT CONFOUNDING BY COHORT

EFFECTS.................................................................................................................................................................. 169

APPENDIX 4.A. CROSS CLASSIFIED RANDOM EFFECTS MODEL (CCREM)....................................... 173

COPYRIGHT ACKNOWLEDGEMENTS ........................................................................................................... 174

ix

List of Tables

TABLE 1.1. STUDY OBJECTIVES AND RESEARCH QUESTIONS ...................................................................................... 4 TABLE 2.1. CHARACTERISTICS OF THE US-STUDIES INCLUDED IN THE META-ANALYSIS BY RACE/ETHNICITY ......... 21 TABLE 2.2. ODDS RATIOS (AND 95% CONFIDENCE INTERVALS) FOR ADVERSE BIRTH OUTCOMES BETWEEN ETHNIC

GROUPS AMONG MIGRANTS, AMONG US-BORN, AND BETWEEN MIGRANTS AND US-BORN, BY

RACE/ETHNICITY. ................................................................................................................................................ 24 TABLE 2.3: CHARACTERISTICS OF THE STUDIES INCLUDED IN THE META-ANALYSIS OF LBW BY WORLD REGIONS 25 TABLE 2.4: ODDS RATIOS (AND 95% CONFIDENCE INTERVALS) FOR LOW BIRTHWEIGHT BETWEEN INFANTS BORN

TO MIGRANT WOMEN FROM VARIOUS WORLD REGIONS IN EUROPE VERSUS EUROPEAN-BORN WOMEN, FROM

VARIOUS WORLD REGIONS IN THE UNITED STATES VERSUS US-BORN WOMEN, AND BETWEEN NATIVE-BORN

AND MIGRANT GROUPS IN EUROPE VERSUS THE US. ....................................................................................... 26 TABLE 3.1: CHARACTERISTICS OF THE STUDY POPULATION, BY MIGRANT STATUS, AND DURATION OF RESIDENCE,

URBAN ONTARIO, LIVE BIRTHS 2002-2007....................................................................................................... 55 TABLE 3.2. UNADJUSTED ODDS RATIOS (AND 95% CONFIDENCE INTERVALS) FOR ADVERSE BIRTH OUTCOMES

BETWEEN NON-IMMIGRANTS AND IMMIGRANTS, OVERALL AND BY DURATION OF RESIDENCE IN CANADA,

2002/2003 TO 2006/2007 ................................................................................................................................ 56 TABLE 3.3: ADJUSTED ODDS RATIOS (AND 95% CONFIDENCE INTERVALS) FOR ADVERSE BIRTH OUTCOMES BY

IMMIGRANTS’ DURATION OF RESIDENCE IN CANADA, URBAN ONTARIO LIVE BIRTHS 2002-2007.................... 58 TABLE 3.4. ODDS RATIOS ADJUSTED FOR ALL COVARIATES IN TABLE 3.3 AND ALSO ADJUSTED FOR PRETERM BIRTH

............................................................................................................................................................................ 58 TABLE 3.5: ADJUSTED* ODDS RATIOS PER 5-YEAR INCREASE IN DURATION OF RESIDENCE IN CANADA (BIRTH TO

IMMIGRANTS 2002-2007), BY WORLD REGION .................................................................................................. 61 TABLE 4.1. CHARACTERISTICS OF THE STUDY POPULATION BY GEOGRAPHY, MEAN INFANT’S BIRTHWEIGHT AND

LOW BIRTHWEIGHT AMONG RECENT IMMIGRANT MOTHERS TO URBAN ONTARIO. ............................................ 85 TABLE 4.2. FIXED EFFECTS (AND 95% CI) OF THE NEIGHBOURHOOD INDICES ON INFANT’S BIRTHWEIGHT (IN

GRAMS) AND RANDOM EFFECTS (AND STANDARD ERRORS) AMONG RECENT IMMIGRANTS TO URBAN ONTARIO.

............................................................................................................................................................................ 87 TABLE 4.3. FIXED EFFECTS (AND 95% CI) OF WORLD REGIONS ON INFANT’S BIRTHWEIGHT (IN GRAMS) AMONG

RECENT IMMIGRANTS TO URBAN ONTARIO. ....................................................................................................... 90 TABLE 4.4. ODDS RATIOS (AND 95% CI) OF WORLD REGIONS ON INFANT’S LOW BIRTHWEIGHT AMONG RECENT

IMMIGRANTS TO URBAN ONTARIO. ..................................................................................................................... 91 TABLE 5.1: ADJUSTED ODDS RATIOS (AND 95% CONFIDENCE INTERVALS) OF ONE STANDARD DEVIATION

INCREASE IN THE MATERIAL DEPRIVATION INDEX AND 10% INCREASE IN THE POPULATION LIVING BELOW THE

LOW-INCOME CUT-OFF ON PRETERM BIRTH, BY IMMIGRANT STATUS AND DURATION OF RESIDENCE, URBAN

ONTARIO, 2002-2007...................................................................................................................................... 110 TABLE 3.A.1. VARIABLE DEFINITIONS........................................................................................................................ 147

x

TABLE 3.B.1. MEASUREMENT OF STILLBIRTHS (FETAL DEATHS PER 1,000 TOTAL BIRTHS) IN THE ONTARIO DAD-

CIHI AT ICES (FISCAL 1988/1989-2006/2007), BY METHOD, AND IN THE ONTARIO VITAL STATISTICS

(CALENDAR 1991-2004) .................................................................................................................................. 152 TABLE 3.B.2. MEASUREMENT OF MULTIPLE BIRTHS (MULTIPLE BIRTHS PER 100 TOTAL BIRTHS) IN THE ONTARIO

DAD-CIHI AT ICES (FISCAL 1988/1989-2006/2007), BY METHOD, AND IN THE ONTARIO VITAL STATISTICS

(CALENDAR 1991-2004) .................................................................................................................................. 155 TABLE 3.B.3. PERCENTILES OF BIRTHWEIGHT (IN GRAMS) BY METHOD .................................................................. 157 TABLE 3.B.4. LIST OF ICD CODES FOR STILLBIRTHS ACCORDING TO REVISION AND UNIT OF ANALYSIS ............... 158 TABLE 3.B.5. LIST OF ICD CODES FOR MULTIPLE BIRTHS ACCORDING TO REVISION AND UNIT OF ANALYSIS ....... 159 TABLE 3.E.1: IMMIGRANTS, BY WORLD REGION, VERSUS NON-IMMIGRANTS........................................................... 170 TABLE 3.E.2. ODDS RATIOS* (AND 85% CONFIDENCE INTERVALS) FOR IMMIGRANTS, BY WORLD REGION, VERSUS

NON-IMMIGRANTS ............................................................................................................................................. 171 TABLE 3.E.3. NUMBER AND PERCENTAGE OF LOW BIRTHWEIGHT (LBW), PRETERM BIRTH (PRT), AND SMALL FOR

GESTATIONAL AGE (SGA), AMONG RESIDENTS OF ONTARIO CENSUS METROPOLITAN AREAS, NON-

IMMIGRANTS AND IMMIGRANTS (ARRIVED 1985-1988) BY DURATION OF RESIDENCE IN CANADA, (BIRTHS

1988/89 TO 2006/2007) ................................................................................................................................. 171 TABLE 3.E.4. ODDS RATIOS* (AND 95% CI) COMPARING IMMIGRANTS BY DURATION OF RESIDENCE VERSUS NON-

IMMIGRANTS...................................................................................................................................................... 172 TABLE 3.E.5. ODDS RATIOS* (AND 95% CI) COMPARING IMMIGRANTS BY DURATION OF RESIDENCE .................. 172

xi

List of Figures

FIGURE 3.1. PREDICTED PROBABILITIES (AND 95% CI) OF PRETERM BIRTH (2002-2007) AMONG ONTARIO

IMMIGRANTS, BY DURATION OF RESIDENCE* ..................................................................................................... 59 FIGURE 3.2. GESTATIONAL AGE DISTRIBUTIONS (AND MEANS) BY DURATION OF RESIDENCE GROUPS.................... 60 FIGURE 3.3. PREDICTED PROBABILITIES* OF PRETERM BIRTH ACCORDING TO DURATION OF RESIDENCE, BY

COHORT OF ARRIVAL .......................................................................................................................................... 62 FIGURE 4.1. DISTRIBUTION OF BIRTHS BY NEIGHBOURHOOD MATERIAL DEPRIVATION TERTILES IN EACH WORLD

SUB-REGION........................................................................................................................................................ 89 FIGURE 4.2: CONFOUNDING BY SELF-SELECTION OF RECENT IMMIGRANTS TO NEIGHBOURHOODS ........................ 96 FIGURE 5.1. PREDICTED PROBABILITIES OF PRETERM BIRTH (2002-2007) BY DURATION OF RESIDENCE AND

NEIGHBOURHOOD DEPRIVATION TERTILES AMONG IMMIGRANTS TO URBAN ONTARIO................................... 111 FIGURE 3.B.1. MEASUREMENT OF STILLBIRTHS IN THE ONTARIO DAD-CIHI AT ICES (FISCAL 1988/1989-

2006/2007), BY METHOD, AND IN THE ONTARIO VITAL STATISTICS (CALENDAR 1991-2004) ..................... 153 FIGURE 3.B.2. MEASUREMENT OF MULTIPLE BIRTHS IN THE ONTARIO DAD-CIHI AT ICES (FISCAL 1988/1989-

2006/2007), BY METHOD, AND IN THE ONTARIO VITAL STATISTICS (CALENDAR 1991-2004) ..................... 156 FIGURE 3.D. DIRECTED ACYCLIC GRAPHS FOR SUFFICIENT CONFOUNDING, BEFORE (3.D.1) AND AFTER (3.D.2)

THE BACKDOOR TEST FOR SUFFICIENCY.......................................................................................................... 164

xii

List of Appendices

APPENDIX 2.A. SEARCH STRATEGY .......................................................................................................................... 140 APPENDIX 3.A. DATA SOURCES................................................................................................................................. 143 APPENDIX 3.B. MEASUREMENT OF STILLBIRTHS AND MULTIPLE BIRTHS USING THE DISCHARGE ABSTRACT

DATABASE ........................................................................................................................................................ 151 APPENDIX 3.C. FLOWCHART DATA EXCLUSIONS .................................................................................................... 162 APPENDIX 3.D. COVARIATE ADJUSTMENT BASED ON DIRECTED ACYCLIC GRAPHS (DAGS).............................. 164 APPENDIX 3.E. USING A COHORT APPROACH TO RULE OUT CONFOUNDING BY COHORT EFFECTS .................... 169 APPENDIX 4.A. CROSS CLASSIFIED RANDOM EFFECTS MODEL (CCREM)............................................................. 173

xiii

Abbreviations

ARTs: Assisted reproductive technologies

BW: Birthweight

CCI: Canadian Classification of Health Interventions

CCP: Canadian Classification of Diagnostic, Therapeutic, and Surgical Procedures

CMA: Census Metropolitan Area

CIC: Citizenship and Immigration Canada

CIHI: Canadian Institute for Health Information

CCREM: Cross-classified Random Effects Model

CRH: Corticotrophin-releasing hormone

CT: Census tract

DA: Dissemination Area

DAD: Discharge Abstract Database

DAG: Directed acyclic graph

EA: Enumeration Area

ICD-9: International Classification of Diseases - 9th Revision

ICD-10-CA: International Classification of Diseases - 10th Revision, enhanced Canadian version

ICES: Institute for Clinical Evaluative Sciences

LBW: Low birthweight

LICO: Statistics Canada low-income cut-off

LIDS: Landed Immigrant Data System

MOHLTC: Ontario Ministry of Health and Long-Term Care

MOOSE: Meta-analysis of Observational Studies in Epidemiology

MLBW: Moderately low birthweight

xiv

MPTB: Moderately preterm birth

OHIP: Ontario Health Insurance Plan

OR: Odds Ratio

PCCF+: Postal Code Conversion File Plus

PPROM: Preterm premature rupture of membranes

PTB: Preterm Birth

REB: Research Ethics Board

ROAM: Reproductive Outcome And Migration

RPDB: Registered Persons Data Base

SAS: Statistical Analysis System

SES: Socioeconomic status

SGA: Small for Gestational Age

SPTB: Small preterm birth

UNICEF: The United Nations Children's Fund

U.S.: United States

VLBW: Very low birthweight

VPTB: Very preterm birth

1

Chapter 1 Introduction

1.1. Background

Immigration has been increasingly seen as a potential means to attenuate population decline,

sustain the size of the labour force, and slow down the pace of population aging.1 As many as

190 million people worldwide are estimated to be international migrants, about half of them

being women.2,3 The proportion of births to migrant women to many industrialized countries has

shown upward trends in the last two decades, reaching around one fifth of all live births in the

US, England and Wales, the Netherlands, Sweden, Switzerland, and Germany.1,4 Despite the

absence of statistics at the national level, Canada does not seem to be the exception. Immigration

currently accounts for two thirds of Canada’s population growth and is expected to be the only

contributing factor by around 2030.5 Although immigrants’ contribution to population growth is

mainly based on the flow of working aged migrants, it is closely followed by that of births to

migrant women, who generally have higher fertility rates than non-immigrants.1

The growing magnitude of the female migrant population and its contribution to Canada’s

newborns give good reason for the study of the health status of immigrants, including

reproductive health. Yet, there are additional important reasons. A second reason is that

immigrants may be different from the ‘native’ population, in terms of their determinants of

health. Indeed, immigrants concentrate a specific set of exposures that are not shared by their

‘native’ counterparts, such as previous environmental exposures in their countries of origin,

selection, language and socialization; these influences shape their health behaviours and status,

and exposures after arrival, such as adaptation and acculturation.6-8 Immigration is not just a

2

single attribute that can be contained into a variable defined at the individual level. Rather,

immigration entails a complex multilevel process that may influence health through diverse

pathways long after settlement and even across generations. This perspective sees immigrants as

a subpopulation with specific determinants of health not shared by the native-born. There is also

some evidence suggesting that common predictors of health outcomes, such as health care

utilization and different measures of socioeconomic status, do not have the same explanatory

power among immigrants than that observed in the general population, as exemplified by the

Mexican paradox, 9,10 by which foreign-born Mexicans show levels of low birthweight similar to

those of the White US-born despite lower prenatal care, education, and income. Immigrant

characteristics associated with enhanced health may not be shared by some immigrant sub-

groups, such as refugees 11 or undocumented immigrants.12

Among these specific determinants we can list the influence of country of origin influences,

adaptation and acculturation, language and culture barriers.

A third and related reason is that the trends of increasing proportions of births to women in a

subpopulation with specific determinants of health about which little is known may have

unexpected consequences for reproductive, child, and adult health. The diversity of the Canadian

immigrant population may contribute to variation in their health outcomes, and such diversity

expedites the detection of effects otherwise undetectable in more homogeneous populations. For

example, despite important advances in our understanding of the etiology of preterm birth during

the last two decades, about one quarter of the occurrence of preterm birth in developed countries

remains unexplained.13 Preterm birth is considered to be a syndrome initiated by multiple

mechanisms, and a precise mechanism cannot be established in most cases.14 Comparisons of

this outcome by migrant status might therefore contribute to our understanding of the

3

environmental influences leading to disparities in preterm birth and associated developmental

and health outcomes.

Despite a substantial body of literature focusing on the reproductive health of migrants to

western industrialised countries, there is no obvious pattern describing the relation between

migrant status and perinatal outcomes. The literature shows conflicting associations between

migration and perinatal health, suggesting different sources of heterogeneity. The heterogeneity

may result from methodological differences between studies, such as the definition of the

migrant groups and the choice of comparison groups as well as differences in the immigrant

populations under study. Unravelling these sources of heterogeneity represents a step towards the

clarification of the relation between migration and perinatal health.

This thesis attempts to contribute to the scientific literature by summarizing the literature on

migration and adverse birth outcomes and tackling some determinants of health specific to

immigrants, such as country of origin influences and duration of residence in the receiving

country, using secondary data on infants born to mothers residing in urban areas of Ontario.

Urban Ontario constitutes an appropriate setting to investigate migrant health. Ontario receives

about half of all immigrants to Canada, with more than 90 % of them concentrated in urban

areas,15 thus representing the largest migrant population of Canada.

4

1.2. Study objectives

Table 1.1. Study objectives and research questions

Objective Research questions Chapter

1. Meta-analysis: To assess disparities by migrants status and migrant subgroups

1.1. Do low birthweight (LBW), preterm birth (PTB), and small for gestational age (SGA) differ between migrants and non-migrants?

1.2. Do these outcomes differ between migrant subgroups, defined by ethnicity?

1.3. Do these outcomes differ between migrant subgroups, defined by country of origin and destination?

2

2. Determinants of birth outcomes for immigrants I: To examine the relation between duration of residence and birth outcomes

2.1. Is duration of residence independently associated with adverse birth outcomes among immigrants to urban Ontario?

2.2. Does such association differ by immigrant subgroups, defined by their regions of origin?

2.3. How do birth outcomes of immigrants, classified by duration of residence, compare with the level observed among non-immigrants?

3

3. Determinants of birth outcomes for immigrants II: To explore the interplay between material deprivation, country of birth, and duration of residence

3.1. What is the contribution of neighbourhood context and country of origin to birth outcomes of recent immigrants?

3.2. Regarding immigrants, is the relation between neighbourhood deprivation and preterm birth modified by duration of residence?

4

3.3. How does the neighbourhood deprivation gradient in preterm birth vary by immigrant status, and by duration of residence among immigrants?

5

5

1.3. Involvement/participation of author and co-authors in the research

Objective 1 arose from my involvement in the ROAM Collaboration (Reproductive Outcomes

and Migration). The ROAM is an international research collaboration between researchers of

Canada, Australia, and several European countries first established in September 2005. It is

being led by Anita Gagnon (McGill University) and was initially funded by the Canadian

Institutes of Health Research (CIHR) under their International Opportunities Program (#157033).

I joined ROAM in mid-2007, under the supervision of Rick Glazier, also a ROAM member. By

that time an overview paper was in preparation,16 the main research question was: “Do migrant

women in ‘western industrialized countries’ have poorer perinatal health outcomes than

receiving-country women (second generation immigrants or non-immigrants)?” A bibliographic

search had been done already for most outcomes (last updated in September 2006) to be used in

the overview paper. In their 2007 annual meeting, ROAM members agreed to split the diversity

of outcomes covered in the overview paper and form working groups to prepare manuscripts

focusing on specific outcomes. Rick Glazier and I agreed to lead a review on migration and low

birth weight and preterm birth. Our working group was composed of 8 other members [Béatrice

Blondel, (INSERM -France), Anita Gagnon (McGill University and MUHC– Canada), Mika

Gissler (STAKES -Finland), Maureen Heaman (University of Manitoba -Canada), Alison

Macfarlane (City University of London – UK), Edward Ng (Statistics Canada), Babill Stray-

Pedersen (University of Oslo – Norway), Jennifer Zeitlin (INSERM – France and EURO-

PERISTAT)]. To avoid overlap with the overview paper, the working groups would preferably

focus on mechanisms. Therefore, Rick and I concentrated on reviewing the literature on low

birth weight among migrant subgroups, and the potential explanations for disparities. ROAM

members provided important comments on the first manuscript plan and their input was critical

in narrowing down the research questions of the review. Once there was agreement on the

6

research questions I took the lead in updating the search (as of November 2007), designing the

study, systematizing the literature, analyzing the data, and writing the manuscripts.

Regarding objectives 2 and 3, I set the objectives, conceived the study design, analyzed and

interpreted the data, and wrote the manuscripts. Committee members supervised the study,

revised and provided critical comments on all manuscripts, including the literature review, from

the protocol to the preliminary results drafts to the final dissertation. They also helped access

data at the Institute of Clinical Evaluative Sciences (ICES) (RG), suggested novel perspectives to

interpret the data (JF), and advised on statistical methods (RM).

1.4. Organization of the thesis

1.4.1. Rationale for the objectives

Objective 1: Literature review:

It is important to review the literature because of the inconsistent associations found between

migration and birth outcomes, more specifically low birthweight, preterm birth, and fetal growth.

The purpose of the review is to clarify the existence and nature of these associations, since a

substantial part of the inconsistencies between studies may be due to differences in

methodological approaches (e.g.., definition of migrant groups, and choice of comparison

groups). A related goal is to learn about the specific determinants of immigrant health and the

mechanisms through which they influence birth outcomes. The knowledge gained from this

review may help avoid common mistakes in the field and fine-tune future research questions and

hypotheses.

7

Objective 2: Duration of residence and birth outcomes:

Heterogeneity between studies may result from unmeasured confounders. Probably one of the

most pervasive and elusive potential confounders or effect modifiers of the associations between

migration and birth outcomes is duration of residence. A few studies found that the association

between duration of residence and adverse birth outcomes is of such a nature that significant

changes in their association could be detected within short periods of time (less than five

years).7,17-19 However, it remains unknown whether such an association is still present when

longer observation periods are considered (up to 20 years). This is a limitation of the literature,

probably due to the lack of available data on these outcomes after prolonged duration of

residence, and because most studies focused on comparisons with the native-born as the referent

groups. Our data made it possible to assess the influence of duration of residence on birth

outcomes of immigrants over a 20-year period after arrival. This feature makes this study not

only original but also very informative for identifying groups at higher risk of adverse birth

outcomes according to their length of stay in Canada. Such information makes the present study

a substantial contribution to the international literature.

Objective 3: Interplay between material deprivation, country of birth, and duration of residence:

One of the most consistent associations in perinatal health is the existence of gradients in

socioeconomic status (SES) (i.e., the lower the SES, the poorer the outcomes in a graded

pattern).20 Immigrants are known to be a subpopulation generally characterized by having lower

SES indicators than in the general population. For example, female Mexicans in the US have

lower maternal education 9 and immigrants to Ontario cities have higher proportions of people

living in deprived neighbourhoods.21 Despite these disadvantages, Mexicans in the US and recent

immigrants to urban Ontario have levels of preterm birth comparable to the most advantaged

8

native-born groups.21,22 Moreover, the seemingly ubiquitous SES gradient in birth outcomes is

not consistently found among these populations.9,21,23,24 These apparent anomalies (or paradoxes)

may be clarified (at least partially) when other determinants of immigrant reproductive health are

considered, such as duration of residence and selection of immigrants into particular sorts of

neighbourhoods, in terms of SES. To assess the impact of SES on birth outcomes of migrants

there are issues of misclassification of immigrants’ SES, confounding by self-selection, and

duration of residence. Objective 3 aims at clarifying the interplay between these dimensions.

Such clarification is important from a public health perspective since it can identify particularly

vulnerable groups defined by the intersection of these dimensions. The contribution of the

present study to the international literature may be significant as well, since U.S. studies trying to

explain the Mexican paradox have not considered duration of residence in their analyses.10,25,26

Objectives 2 and 3 build upon my MSc thesis work that resulted in two papers: one reporting

lower twin rates among recent immigrants and residents of poorer neighbourhoods,27 and the

other reporting lower singleton preterm birth rates among recent immigrants but higher low-

birthweight rates.21 However, this research had two major limitations: first, recent immigration

was ascertained by means of a proxy (defined as a first-time registration to OHIP within a five-

year period preceding the delivery) and this proxy has not been validated; second, information on

the maternal country of origin was not available, which made it impossible to identify variations

by immigrant sub-groups. In this PhD thesis I tried to overcome these two limitations and

advance knowledge on the relation between immigration and birth outcomes using the Landed

Immigrant Data System (LIDS), which is the official immigration database and contains mostly

validated information on immigrant status, country of birth and origin (last permanent residence),

and date of arrival, among other characteristics as described in the Appendix on data sources

(Appendix 3.A). Access to the LIDS and linkage with the Discharge Abstract Database

9

(DAD),28,29 compiled by the Canadian Institute for Health Information (CIHI), allowed two types

of comparisons: 1) immigrants versus the rest of the population; and 2) comparisons between

immigrant subgroups, defined by world regions of origin and length of stay in Canada. Both

types of comparisons have been found relevant from an equity perspective. First, the general

population constitutes a referent group against which to compare immigrants as a whole and

assess the impact of immigration at the societal level. Secondly, to study differences between

immigrant sub-groups is of interest to identify high-risk sub-populations, understand the

underlying determinants of their health status, and inform potential strategies to improve their

outcomes.

1.4.2. Overview of the research

This section highlights the location of the study objectives in the thesis. Birth outcomes of

immigrants constitute the common theme of the following four papers. Although the focus was

on immigrants, comparisons with non-immigrants were made whenever feasible and relevant.

This thesis is limited to a few birth outcomes such as birthweight, preterm birth and small for

gestational age, which are intermediate outcomes in perinatology. Other potentially more

relevant outcomes such as stillbirth, neonatal mortality, and serious neonatal morbidity were not

chosen due to data limitations. Papers are ordered according to the increasing complexity of their

objectives.

The literature review (Chapter 2) was of great importance in classifying the literature, thus

providing a context for the interpretation of the Canadian data analyzed in Chapters 3 to 5. It also

helped identify what is known and not known on this topic, thus paving the way to the

10

recognition of two research gaps that could be explored with the Ontario data analyzed in

objectives 2 and 3.

The first research gap identified in the review was the relation between duration of residence and

adverse birth outcomes (Chapter 3). The literature review identified duration of residence as an

important determinant. Yet, the very few studies that have been able to collect data on duration

of residence were limited to recent immigrants.7,17,18 In contrast, chapter 3 is unique in assessing

the effects of the receiving environment across the 20 years after migration. The main finding

reported in this chapter is that duration of residence was linearly associated with higher odds of

low birthweight and preterm birth, most probably due to a steady decrease in the gestational age

distribution of immigrants with time spent in Canada.

The second gap is the relation between SES and birth outcomes among migrants. Migrants to

western industrialized countries represent a major exception to the well-known negative

association between SES and adverse birth outcomes. Chapter 4 provides an explanation of why

this association does not hold among recent immigrants to urban Ontario. These analyses were

carried out during 2007 and are restricted to infants born to recent immigrants mothers from

1993 to 2001. Additional data were accessed in 2008, thus allowing us to assess the effects of

duration of residence on birth outcomes of immigrants. Therefore, Chapter 5 goes beyond recent

immigrants and extends the observation period by including immigrants with 20 or more years of

stay in Canada. These data allowed the assessment of whether the influence of neighbourhood

deprivation on preterm birth was modified by immigrants’ duration of residence, and how these

gradients compared with that of the non-immigrant population.

11

The final discussion (Chapter 6) highlights the general strengths and limitations of the thesis,

summarizes the lessons learnt, and signals possible directions to further advance knowledge in

the area of migration and adverse birth outcomes.

12

1.5. References

(1) Sobotka T. Overview Chapter 7: The rising importance of migrants for childbearing in Europe.

Demographic Research. 2008;19(article 9):225-248.

(2) Carballo M, Nerukar A. Migration, refugees, and health risks. Emerging Infectious Diseases. 2001;7(3

Suppl):556-560.

(3) United Nations Population Fund (UNFPA). State of World Population 2006. A Passage to Hope: Women

and International Migration. 1-116. 2006. New York, UNFPA.

(4) Centers for Disease Control and Prevention (CDC), Centers for Disease Control and Prevention (CDC).

State-specific trends in U.S. live births to women born outside the 50 states and the District of Columbia--

United States, 1990 and 2000.[erratum appears in MMWR Morb Mortal Wkly Rep. 2002 Dec

13;51(49):1127.]. MMWR - Morbidity & Mortality Weekly Report. 2002;51(48):1091-1095.

(5) Bélanger, A, Martel, L, and Caron-Malenfant, E. Population Projections for Canada, Provinces and

Territories 2005-2031. Statistics Canada. Catalogue no. 91-520-XIE, 1-215. 2005. Ottawa, Statistics

Canada. Demography Division.

(6) English PB, Kharrazi M, Guendelman S. Pregnancy outcomes and risk factors in Mexican Americans: the

effect of language use and mother's birthplace. Ethnicity & Disease. 1997;7(3):229-240.

(7) Guendelman S, English PB. Effect of United States residence on birth outcomes among Mexican

immigrants: an exploratory study. Am J Epid. 1995;142(9 Suppl):S30-S38.

(8) Scribner R, Dwyer JH. Acculturation and low birthweight among Latinos in the Hispanic HANES. Am J

Public Health. 1989;79(9):1263-1267.

(9) Acevedo-Garcia D, Soobader MJ, Berkman LF. The differential effect of foreign-born status on low birth

weight by race/ethnicity and education. Pediatrics. 2005;115(1):e20-e30.

(10) Rosenberg TJ, Raggio TP, Chiasson MA. A further examination of the "epidemiologic paradox": birth

outcomes among Latinas. Journal of the National Medical Association. 2005;97(4):550-556.

13

(11) Desmeules M, Gold J, Kazanjian A et al. New approaches to immigrant health assessment. Can J Public

Health. 2004;95(3):I22-I26.

(12) Kelaher M, Jessop DJ. Differences in low-birthweight among documented and undocumented foreign-born

and US-born Latinas. SOC SCI MED. 2002;55(12):2171-2175.

(13) Kramer MS, Seguin L, Lydon J, Goulet L. Socio-economic disparities in pregnancy outcome: why do the

poor fare so poorly? Paediatr Perinat Epidemiol. 2000;14(3):194-210.

(14) Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet.

2008;371(9606):75-84.

(15) Statistics Canada. Immigration in Canada: A Portrait of the Foreign-born. Census year 2006. Catalogue no.

97-557-XIE. 2007. Ottawa, Statistics Canada, Minister of Industry.

(16) Gagnon, A. J., Zimbeck, M., Zeitlin, J., and and the ROAM Collaboration. Migration to western

industrialized countries and perinatal health: A systematic review. Social Science & Medicine (in press).

2009.

(17) Crump C, Lipsky S, Mueller BA. Adverse birth outcomes among Mexican-Americans: are US-born women

at greater risk than Mexico-born women? Ethnicity & Health. 1999;4(1-2):29-34.

(18) Rasmussen F, Oldenburg CE, Ericson A, Gunnarskog J. Preterm birth and low birthweight among children

of Swedish and immigrant women between 1978 and 1990.[erratum appears in Paediatr Perinat Epidemiol

1996 Apr;10(2):240-1]. Paediatr Peri Epid. 1995;9(4):441-454.

(19) Ray JG, Vermeulen MJ, Schull MJ, Singh G, Shah R, Redelmeier DA. Results of the Recent Immigrant

Pregnancy and Perinatal Long-term Evaluation Study (RIPPLES). CMAJ. 2007;176(10):1419-1426.



(21) Urquia ML, Frank JW, Glazier RH, Moineddin R. Birth outcomes by neighbourhood income and recent

immigration in Toronto. Health Rep. 2007;18(4):1-10.

14

(22) Cervantes A, Keith L, Wyshak G. Adverse birth outcomes among native-born and immigrant women:

replicating national evidence regarding Mexicans at the local level. Matern Child Health J. 1999;3(2):99-

109.

(23) Fang J, Madhavan S, Alderman MH. Low birth weight: race and maternal nativity-impact of community

income. Pediatrics. 1999;103(1):E5.

(24) Pearl M, Braveman P, Abrams B. The relationship of neighbourhood socioeconomic characteristics to

birthweight among 5 ethnic groups in California. Am J Public Health. 2001;91(11):1808-1814.

(25) Buekens P, Notzon F, Kotelchuck M, Wilcox A. Why do Mexican Americans give birth to few low-birth-

weight infants? Am J Epid. 2000;152(4):347-351.

(26) Fuentes-Afflick E, Hessol NA, Perez-Stable EJ. Testing the epidemiologic paradox of low birth weight in

Latinos. Archives of Pediatrics & Adolescent Medicine. 1999;153(2):147-153.

(27) Urquia ML, Frank JW, Glazier RH, Moineddin R. Multiple maternities and neighbourhood income. Twin

Res Hum Genet. 2007;10(2):400-405.

(28) Canadian Institute for Health Information. Data Quality Documentation: Discharge Abstract Database

2001–2002. 2003. Ottawa, Canadian Institute for Health Information.



15

Chapter 2 International Migration and Adverse Birth Outcomes: Role of

Ethnicity, Region of Origin and Destination

Abstract

Objective: To determine whether low birthweight (LBW), preterm birth (PTB), and small for

gestational age (SGA) differ between non-migrants and migrant subgroups, defined by

race/ethnicity and world region of origin and destination.

Methods: We conducted a systematic review and meta-analysis using three-level logistic

regression models to account for the heterogeneity between studies and between subgroups

within studies.

Results: Twenty six studies, involving more 30 million singleton births, met inclusion criteria.

Compared to US-born Black women, first-generation Black migrant women had lower odds of

delivering LBW (OR, 95% confidence intervals: 0.64, 0.50-0.83), PTB (0.70, 0.62-0.80) and

SGA babies (0.65, 0.49-0.85). Hispanic migrants also exhibited lower odds for these outcomes,

but Asian and White migrants did not. Sub-Saharan African and Latin American and Caribbean

women were at higher odds of delivering LBW babies in Europe but not in the US and South

Asians were at higher odds in both continents, compared with the native-born populations.

Conclusions: The association between migration and adverse birth outcomes varies by migrant

subgroup and it is sensitive to the definition of the migrant and reference groups.

16

2.1. Introduction

About 95 million women are international migrants worldwide and female immigrants have

recently outnumbered male immigrants in most industrialised countries.1 The proportion of live

births to immigrant women has risen during recent decades in several industrialized countries.2-5

Despite a substantial body of literature focusing on the reproductive health of migrants to

western industrialised countries, there is no obvious pattern describing the relation between

migrant status and perinatal outcomes. The literature shows positive, negative, and null

associations between migration and perinatal health, suggesting that different sources of

heterogeneity may play a role. It is uncertain to what extent the association between foreign-born

status and birth outcomes is a function of the characteristics of the migrant populations, of the

baseline risk of the native-born reference groups, or of some combination of both. For example,

foreign-born Blacks in the United States (US) had lower low birthweight compared with US-

born Blacks but not with US-born Whites.6 Such comparisons suggest that the influence of

migration may be modified by ethnicity.7 Ethnic disparities in birth outcomes are well

documented, particularly in the US, but the contribution of migration to these disparities is not

well understood. In studies comparing native-born versus migrant groups defined by their

regions of origin, there is uncertainty over whether the so-called healthy migrant effect 8 applies

to migrants from all or only some regions of the world, and what these regions are.

Most studies devoted to migration and perinatal health have focused on birth outcomes defined

by birthweight or gestational age or both. Our purpose was to conduct a systematic review to

clarify the relation between migration and those birth outcomes by determining the differences in

low birthweight (LBW), preterm birth (PTB), and small for gestational age (SGA) between

17

migrants and non-migrants by migrant subgroups, defined according to race/ethnicity, world

region of origin and actual destination.

2.2. Methods

This review was prepared following the MOOSE guidelines9 and draws on the material

identified by the ROAM collaboration (Reproductive Outcome And Migration) for a series of

systematic reviews on migration and reproductive health. The ROAM is an international research

collaboration devoted to study the relation between reproductive health and migration.

2.2.1. Study population

This study was restricted to published reports on any outcome requiring gestational age or

infant’s birthweight to define it. The exposure was maternal international migration to Western

industrialised countries, assessed by evidence of cross-border movement. Thus this definition

excludes internal migration, ‘protectorates’ such as Puerto Rico, and second generation

populations. Reference groups were the native-born women of the receiving countries. We

excluded case studies, clinical reports, reports without a comparison group, and reports in which

the results of the migrant group(s) were not presented separately from the comparison group.

2.2.2. Search and study selection criteria

Studies were identified through electronic literature databases from 1995 through October 2007

to represent the most recent immigration context. Ovid (version 10.5.1) was used in the

following order: Medline, Health Star, Embase, and PsychInfo. Searches were supplemented

with bibliographic citation hand-searches of included articles published from 2004 onward and

18

relevant articles referred to the authors by other ROAM members. No language exclusions were

routinely applied. Articles in French, Italian, and Spanish, were reviewed by the authors. Two

ROAM members independently assessed included studies for quality using the US Preventative

Services Task Force criteria for cohort and case-control studies 10 and no discrepancies were

found in the overall score between raters. Further details of the search strategy are summarized in

Appendix 2.A.

All articles for the meta-analyses were selected by applying the following criteria:

1. Definitions of the outcomes: LBW was restricted to a birthweight less than 2500 grams,

PTB to a gestational age of less than 37 completed weeks, but no restrictions were

applied to SGA, due to the small number of studies and variations in the definitions.

2. Restriction to singleton births.

3. Information on race/ethnicity and foreign-born status or country of birth or nationality

4. Descriptive tables including summary data on the outcomes with at least one native-born

and one foreign-born group.

Our searches identified eighty two studies. Of these, we excluded 10 studies from the meta-

analyses that did not include our outcomes of interest or used different definitions,11-20 31 studies

that did not discriminate between singleton and multiple births,2,3,5,21-48 four did not ascertain

migration appropriately,49-52 and seven did not have appropriate tables for the extraction of the

data.53-58 Finally, four studies 59-62 could not be used because the available number of studies

reporting PTB and SGA by world region of origin was too small for analysis. Twenty six studies

were meta-analyzed.

19

2.2.3. Data extraction

We extracted summary birth data consisting of at least two records per study: one for the migrant

and one for the native-born group, although many studies included several subgroups including

maternal ethnic groups, world regions or countries of origin or infant’s year of birth. Each record

contained a numerator and a denominator for the respective outcome, and indicators of migrant

status (foreign-born, native-born), race/ethnicity as categorised in US studies (Asians, Blacks,

Hispanics, and Whites),63 receiving country (US or European countries), migrants’ country of

birth or origin or nationality, and infant’s year of birth. If the birth data aggregated more than one

year, the midpoint was recorded, and for articles reporting numerator and denominator for

different periods, one record was assigned to each period. We grouped countries of birth into

world regions, following the classification of the United Nations in most cases.64 Asia was

subdivided into South Asia and rest of Asia, because we wanted to examine whether South

Asians differ in the risk of adverse birth outcomes compared to the rest of the continent, as has

been suggested in the literature.65 In the same vein, North Africans were separated from the rest

of Africa (i.e., Sub-Saharan Africa) because of their particularly good birth outcomes,66 and

grouped with Middle Eastern countries, because some studies 60,67 have grouped these regions

together. Sensitivity analyses performed without these two studies did not affect the results

regarding North Africans and therefore we did not exclude them.

2.2.4. Statistical analyses

In order to account for the potential heterogeneity between studies and subgroups within studies,

we employed random effects meta-regression analysis, which involves the application of

multilevel methods to meta-analysis.68-70 We used three-level models, with births at level 1,

20

subgroups at level 2 and studies at level 3. The inclusion of random effects at the subgroup-level

assumes that each subgroup represents a different population with its own distribution. Ignoring

the hierarchical structure of these data would produce over-precise confidence intervals.71,72

Analyses were conducted with Proc GLIMMIX in SAS 9.1 (SAS Institute, Cary, NC) to fit

multilevel logistic regression models for summary data.

Studies differed substantially in the way migrant groups were categorised. This heterogeneity in

the definition of migrant groups prevented us from combining all selected studies into one single

meta-analysis and therefore we conducted two meta-analyses, based on the two main approaches

that have been used to study the influences of international migration on birth outcomes.

The first meta-analysis was based on studies which analyzed births in the US defining migrant

subgroups both by their race/ethnicity 63 and migrant status, because most European studies did

not report birth data by ethnic groups. We used a product term between race/ethnicity and

migrant status in the models to obtain effect estimates of migrant status by race/ethnicity on all

three outcomes, adjusted for infant’s year of birth. We quantified the percent of variance

explained for logistic models by comparing a model including the product term between

race/ethnicity and migrant status relative to a model including migrant status as the only

predictor, with both models adjusted for infant’s year of birth.73

The second meta-analysis was based on studies which analyzed low birthweight in Europe or the

US, categorizing migrants and non-migrants by their countries of birth, irrespective of their

race/ethnicity. We could not analyze preterm birth and small for gestational age due to the small

number of studies and migrant groups. The low birthweight model included a product term

between world region of origin and a dummy indicator for receiving country (Europe versus US)

21

in order to test the hypothesis that the odds of LBW differ both according to the region of origin

and destination, adjusted for infant’s year of birth.

2.3. Results

Twenty six studies were included in the meta-analyses: 17 by race/ethnicity,6-8,74-87 17 by world

region,6,8,13,66,67,77,79,81,82,84,85,87-92 and 9 by both.6,8,77,79,81,82,84,85,87 Due to the small number of

studies it was not possible to choose a uniform definition of SGA, and therefore all were

considered, including SGA based on percentiles of the birth weight distribution of native-born

populations 8,76,81,85,91, full-term LBW infants,77,84 and other measures of SGA.83

Table 2.1. Characteristics of the US-studies included in the meta-analysis by race/ethnicity

Study (author, year)

Country, state/region

Type of databas

e b

Year of data

Outcome Migrants US-born # of subgroups

Total births a % migrant

Acevedo-Garcia et al. 2005

USA, national

PBR 1998 LBW Asians, Blacks, Whites

Asians, Blacks, Whites

6 2,102,393 9.3

Alexander et al. 1996

USA, regional NE

PBR 1983-1987 LBW Asians Asians 2 37,941 45.3

Cervantes et al. 1999

USA, Chicago City

PBR 1994 LBW, PTB Blacks, Hispanics,

Whites

Blacks, Hispanics,

Whites

8 52,033 27.0

Cocroft et al. 2002

USA, New York State

PBR 1993-1996 SGA Blacks, Hispanics,

Whites

Blacks, Hispanics,

Whites

8 2,356 21.2

Crump et al. 1999

USA, Washington State

PBR 1989-1994 LBW, PTB, SGA

Hispanics Hispanics 2 9,572 50.0

David et al. 1997

USA, Illinois State

PBR 1980-1995 LBW Blacks Blacks, Whites 3 90,503 3.5

English et al. 1997

USA, California

PBR + quest

1992 LBW, PTB Hispanics Hispanics 6 4,404 55.3

Fang et al. 1999

USA, New York City

PBR 1988-1994 LBW, PTB Blacks Blacks 5 269,863 35.9

Fuentes-Afflick et al. 1998

USA, California State

PBR 1992 LBW, PTB Asians, Blacks, Hispanics,

Whites

Asians, Blacks, Hispanics,

Whites

8 573,233 44.5

Gould et al. 2003



Asians, Hispanics

Blacks, Whites 4 1,057,977 42.2

Johnson et al. 2005


PBR 1993-2001 LBW, PTB Blacks Blacks, Whites 3 5,398 10.7

Kramer et al. USA, PBR 1998-2000 PTB, SGA Blacks Blacks 2 1,754,777 11.4

22

Study (author, year)

Country, state/region

Type of databas

e b

Year of data

Outcome Migrants US-born # of subgroups

Total births a % migrant

2006 national Landale et al. 1999

USA, national

PBR 1989-1991 LBW, SGA Asians, Blacks, Hispanics,

Whites

Asians, Blacks, Hispanics,

Whites

36 4,856,798 48.6

Madan et al. 2006

11 States PBR LBW, PTB, SGA

Asians, Hispanics,

Asians, Hispanics,

Whites

5 6,424,172 23.1

California, Hawaii, Illinois, New Jersey, New York, Texas, Washington

1995-1997

Minnesota 1997 Virginia 1998 Missouri,

West Virginia 1999-2000

Palotto et al. 2000

USA, Illinois State

PBR 1985-1990 LBW Blacks Blacks, Whites 3 103,746 2.2

Rosenberg et al. 2005

USA, New York City

PBR 1996-1997 LBW Hispanics Hispanics 14 156,084 63.1

Wingate et al. 2006

USA, national


Hispanics Hispanics 4 2,446,253 61.5

TOTAL 119 19,947,503 a When the sample size varies by outcome, the denominator for LBW was reported, followed by PTB and SGA if LBW was not reported. b HR: hospital record, PB: population-based, PBR: population-based registry, PBS: population-based survey

2.3.1. Migration and race/ethnicity

The first meta-analysis was based on 17 studies conducted in the US (Table 2.1). All studies used

self-reported race/ethnicity and foreign-born status. One UK study 13 also reported these data for

LBW but was excluded to restrict our analysis to the US context. Exclusion of this study did not

substantially change the effect estimates for LBW although it slightly increased the p-value for

the product term between race/ethnicity and migrant status shown below. We also excluded

Hispanics from one US study 7 to avoid data duplication with another study.8

We first fitted a three-level model with migrant status as the independent variable, adjusted for

infant’s year of birth, but ignoring race/ethnicity. The odds ratios (95% confidence interval) for

the comparisons between migrants and non-migrants were 0.81 (0.70-0.94) for LBW, 0.85 (0.74-

0.98) for PTB and 0.89 (0.72-1.10) for SGA, respectively. These are inappropriate models that

23

assume that the effect of migrant status can be averaged across racial/ethnic groups. Instead,

Table 2.2 shows the results of the three-level models including race/ethnicity and a product term

between race/ethnicity and migrant status for all three outcomes, adjusted for year of birth. The

p-values of the product term in the models were 0.0611 for LBW, 0.0018 for PTB, and 0.0013

for SGA. The percent of total variance explained by the introduction of race/ethnicity and the

product term “migrant status * race/ethnicity” relative to a model including only migrant status,

adjusted for year of birth, was 57%, 24% and 26% for LBW, PTB, and SGA, respectively,

suggesting that race/ethnicity and its interplay with migrant status explain substantial variability

in the outcomes not accounted for by migrant status alone.

The first, second and third columns of Table 2.2 present ethnic disparities within first generation

migrants, within US-born, and between foreign-born and US-born of the same race/ethnicity,

respectively, by outcome. The percent of singleton infants with LBW was 5%, 9% for PTB, and

3% for SGA among US-born White mothers. There was no evidence of differences in the

outcomes between foreign-born and US-born Whites (third column).

Among foreign-born migrants, Asians were somewhat more likely to have adverse birth

outcomes than Whites. Among the US-born, Asians were at higher odds of SGA compared with

Whites. Foreign-born Asians were not protected against these outcomes, compared to US-born

Asians.

24

Table 2.2. Odds ratios (and 95% confidence intervals) for adverse birth outcomes between ethnic groups

among migrants, among US-born, and between migrants and US-born, by race/ethnicity.

Outcome Ethnicity Migrants US-born Migrants versus US-bornb Low birthweight N=6,487,938 N=11,702,432 OR (95% CI) a OR (95% CI) a OR (95% CI) a Whites 1.00 1.00 0.83 (0.60-1.16) Asians 1.31 (0.95-1.80) 1.14 (0.88-1.48) 0.95 (0.75-1.21) Blacks 2.41 (1.73-3.35) 3.10 (2.38-4.04) 0.64 (0.50-0.83) Hispanics 1.16 (0.86-1.57) 1.25 (1.00-1.58) 0.77 (0.64-0.93) Preterm birth N=4,009,158 N=8,587,564 OR (95% CI) a OR (95% CI) a OR (95% CI) a Whites 1.00 1.00 0.82 (0.66-1.01) Asians 1.44 (1.15-1.81) 1.08 (0.88-1.35) 1.09 (0.88-1.35) Blacks 1.62 (1.30-2.03) 1.89 (1.64-2.19) 0.70 (0.62-0.80) Hispanics 1.35 (1.10-1.66) 1.24 (1.09-1.44) 0.89 (0.79-1.00) Small for Gestational age

N=5,268,854 N=9,299,526

OR (95% CI) a OR (95% CI) a OR (95% CI) a Whites 1.00 1.00 1.13 (0.78-1.63) Asians 1.67 (1.18-2.36) 1.63 (1.28-2.08) 1.15 (0.94-1.41) Blacks 1.48 (0.99-2.21) 2.58 (2.00-3.34) 0.65 (0.49-0.85) Hispanics 0.97 (0.69-1.37) 1.34 (1.07-1.69) 0.81 (0.68-0.97)

a Includes random effects at the subgroup and study levels. Adjusted for infant’s year of birth b US-born is the reference group

Among the foreign-born, Black migrants were the group at the highest odds for LBW and PTB,

and Blacks were the group at the highest odds for all three outcomes among US-born women.

There was evidence of a protective effect of being foreign-born among Blacks for all three

outcomes.

Foreign-born Hispanics were at slightly higher odds for PTB compared with foreign-born White

migrants. The Hispanic-White gap was wider among the native-born than among the foreign-

born in LBW and SGA but not in PTB. Hispanic migrants were at lower odds of all three

outcomes compared with their US-born counterparts.

25

2.3.2. Migration and world regions

Table 2.3: Characteristics of the studies included in the meta-analysis of LBW by World regions

Study Country, state/region

Type of database b

Year of data

Migrants’ world regions # of subgroups

Total births % migrants

Buekens/ 1998

Belgium, national

PBR 1981-1988 North Africa 2 839 972 4.2

Collinwood-Bakeo/ 2004

UK, national PBR 1983-2001 Caribbean, East Africa, West Africa, South Asia, East Europe, Western Europe

55 11 401 247 8.0

Crump/ 1999 USA, Washington State

PBR 1989-1994 Latin America (Mexico) 2 9 572 50.0

David/ 1997 USA, Illinois State

PBR 1980-1995 Sub-Saharan Africa 3 90 503 3.5

Fang/ 1999 USA, New York City

PBR 1988-1994 Caribbean, South America, Africa (excl North)

5 269 863 35.9

Fuentes-Afflick/ 1997


PBR 1992 Cambodia, China, Philippines, India, Korea, Laos, Thailand, Vietnam

9 268 949 17.5

Gissler/ 2003 Sweden, national

PBR 1987-1988 Finland 6 140 390 23.8

Gould/ 2003 USA, California

PBR 1995-1997 India, Mexico 4 1 057 977 42.2

Guendelman/ 1999

Belgium, national

PBR 1992 North Africa 2 107 968 4.3

France, national

PBS 1995 North Africa 2 11 802 5.4

USA, national

PBR 1995 Latin America (Mexico) 2 3 417 003 8.4

Harding/ 2004

UK, national PBR 1983-2000 South Asia, Caribbean, Africa (excl North)

11 57 474 6.3

Johnson/ 2005


PBR 1993-2001 Somalia 3 5 398 10.7

Landale/ 1999

USA, national

PBR 1989-1991 Latin America, China, Philippines, Japan

16 4 856 798 48.6

Madan/ 2006 USA, national

PBR 1995-2000 India, Latin America (Mexico)

5 6 424 172 23.1

Rasmussen/ 1995

Sweden, national

PBR 1978-1990 West Europe/North America, East Europe, North Africa/Middle East, Sub-Saharan Africa, Latin America,

8 1 265 531 11.8

Rosenberg/ 2005

USA, New York City

PBR 1996-1997 Latin America 12 156 084 63.1

Vangen/ 2002

Norway, national

PBR 1980-1995 Pakistan, Vietnam, North Africa

4 820 256 1.4

Wingate/ 2006

USA, national

PBR 1995-1999 Latin America (Mexico) 4 2 446 253 61.5

TOTAL 155 33 647 212 b HR: hospital record, PB: population-based, PBR: population-based registry, PBS: population-based survey

26

Table 2.4: Odds ratios (and 95% confidence intervals) for low birthweight between infants born to migrant

women from various World Regions in Europe versus European-born women, from various World

Regions in the United States versus US-born women, and between native-born and migrant groups in

Europe versus the US.

Infants born in Europe

Infants born in the US

Infants born in Europe versus in the

USb OR (95% CI) a OR (95% CI) a OR (95% CI) a

Native-born women 1.00 1.00 0.66 (0.50-0.86)

Migrants from:

Western Europe and North America 0.84 (0.72-0.99) - -

East Europe 0.91 (0.73-1.14) - -

North Africa/ME 0.75 (0.58-0.98) - -

Sub-Saharan Africa 1.57 (1.31-1.89) 0.74 (0.54-1.01) 1.40 (0.93-2.11)

South Asia 1.66 (1.41-1.94) 1.25 (1.01-1.55) 0.87 (0.62-1.22)

Rest of Asia - 0.78 (0.62-0.99) -

Latin America / Caribbean 1.32 (1.06-1.64) 0.72 (0.62-0.82) 1.21 (0.88-1.68) a Includes random effects at the subgroup and study levels. Adjusted for infant’s year of birth. b Infants born in the US are the reference group

The second meta-analysis was based on studies that ascertained migration by the maternal

country of birth or nationality and included several European countries (Table 2.3). In one study

that stratified the outcomes by national origin we included national-origin groups with at least

90% of foreign-born women and therefore excluded Japanese and Filipino women (25).

Table 2.4 presents the results of the three-level model including a product term between world

region of origin and destination (Europe versus US) (p-value <0.0001). A few comparisons were

not possible because some subgroups migrating to Europe or to the US were not represented in

the selected studies.

Women from Sub-Saharan Africa and Latin America and the Caribbean were at higher odds for

LBW if migrating to European countries but at lower odds if migrating to the US, compared to

27

the respective native-born women. South Asians were at higher odds in both contexts but the

association was stronger in Europe. The direction and strength of these associations are affected

by the different baseline risk of the European and US reference groups, with European-born

women less likely to deliver LBW infants compared to US-born women (OR = 0.66 [95% CI =

0.50-086]). Despite this, Sub Saharan African women migrating to Europe seemed to be more

likely to deliver LBW babies compared to those from the same region who migrated to the US,

although there was no strong evidence to support the hypothesis that LBW within migrant groups

differed according to their destination (third column).

2.4. Discussion

One of the main findings of this systematic review is that the association between foreign-born

status and birth outcomes is not uniform but depends on the migrant subgroup, either defined by

a combination of maternal race/ethnicity and migrant status or by the world region of origin and

actual destination. We found that infants born to first-generation Black and Hispanic migrant

women were at lower risk of adverse birth outcomes than their US-born counterparts, but did not

find evidence of such protective effect among Asians and Whites. Migrants from these

ethnicities were at higher risk than White migrants overall. Regarding subgroups defined by

region of origin, Sub-Saharan African and Latin American and Caribbean migrants were at

higher odds of LBW in Europe but not in the US and South Asians were at higher odds in both

continents.

Unlike most meta-analyses of observational studies, we chose to use unadjusted summary data.

This approach had the advantage of freeing comparison groups from the reported measures of

28

effect in multivariate analyses thus making it possible to examine comparisons not explored in

previous studies. This allowed assessment of ethnic disparities by migrant status and

comparisons within migrant groups according to their place of origin and destination. Another

advantage is that our analyses used the same set of covariates and definitions for each study

(with the exception of SGA) thus making interpretation of results less problematic than in meta-

analyses based on effect estimates adjusted for varying number of covariates with heterogeneous

definitions. However, the limit of our approach was the inability to extract birth data stratified by

potential confounders. Immigration policies in the receiving countries and social class dynamics

in the source countries may favour the selection of women or couples for migration, based on

certain characteristics for which distributions may differ both from those of the source and the

receiving population (e.g., maternal age, maternal and paternal social class, marital status, overall

health) and that are also associated with birth outcomes. For example, differences in maternal

age may explain part of the foreign-born advantage among Blacks and Hispanics in the US, since

these groups have lower proportions of teenage pregnancy than their US-born

counterparts.7,75,79,80,84 Those two groups had also lower proportions of single mothers.7,75,80

Despite these favourable characteristics foreign-born Mexicans but not foreign-born Blacks in

the US had lower education, less prenatal care, and lower income compared to US-born

mothers.7,77,85 This phenomenon makes up part of the so-called “Latino paradox”,52,54,84,87 that

also can be extended to the birthweight advantage of North Africans in France and Belgium.12,93

The appropriateness of adjusting for factors likely affected by exposure, such as prenatal care

use, language knowledge, and health behaviours, mainly tobacco smoking, alcohol consumption

and drug use, is dubious, although this has been done in several studies.7,8,76,78-80

Another potential source of bias is measurement error, mainly resulting from self-reported

race/ethnicity and country of birth and nationality in birth certificates. A validation study found

29

that at least 94% of the mothers from the major racial/ethnic groups whose race/ethnicity was

reported in 1994-1995 California birth certificates were also classified in the same way in a face-

to-face structured postpartum interview.94 Similar results were obtained in a previous national

study,95 suggesting that the impact of this bias is small. The meaning and limitations of the

racial/ethnic classification for epidemiologic research had been extensively discussed.96,97 The

reviewed literature on birth outcomes tended to consider the racial/ethnic categories as markers

for a social process external to individual physiology rather than indicators of biological types.

Although reduced by means of model specification, heterogeneity across studies and subgroups

remained in our full models. Unexplained variability may result from unmeasured characteristics

and from the residual variability between countries within migrant subgroups. Some effect

estimates may also be affected by the uneven representation of some countries within world

regions between receiving countries, such as those of the Latin America and the Caribbean

subgroup, mainly driven by Mexicans in the US but not in Europe. In the same vein, residual

confounding may also be due to different distributions of generations within ethnic groups, with

US-born Hispanics and Asians more likely to be second generation than US-born Blacks or

Whites, who are mostly fourth or higher generation.98 Even first generation migrants may differ

in their risk of adverse birth outcomes according to their length of residence in the receiving

country, information that was rarely collected.51,67,77

Studies were heterogeneous in the way migrants were grouped into world regions, possibly

reflecting the diversity of immigration patterns between receiving countries, lack of consensus

about appropriate classifications, constraints in the availability of data, or population size

considerations. We did not assess publication bias because selected studies were mostly

exploratory, had high variation in their objectives and hypotheses and in the choice of the

30

comparison groups. However, publication bias may be present if some authors were unwilling or

unable to publish studies showing no associations between migration and pregnancy outcomes,

although the direction of such expected associations is not obvious, given the diversity of

migrant and native-born groups.

Our findings should be regarded as global tendencies that may not apply to particular migrant

subgroups settling in particular countries, regions or even cities. Although it is recognised that

LBW may result from either early delivery or fetal growth restriction,99 explaining results of

LBW based on the results of PTB and SGA for particular groups should be limited to hypothesis

generation, since the studies included in the analyses are not the same for LBW, PTB and SGA.

Differences in birth outcomes by migrant status may be influenced by the processes of selective

migration and acculturation. Selective migration has been one of the major mechanisms used to

explain the birth weight advantage of migrant compared with receiving-country women,

particularly Mexicans in the US,6,7,18,29,66,74,79 although there was no direct evidence to support

this hypothesis. While some studies referred to the ‘healthy migrant effect’ by highlighting the

favourable distribution of some risk factors for birth outcomes among migrant versus non-

migrant women,6,26,33 the relevant comparison group for assessing selective migration is the one

that stayed in the home countries100 and the few studies referring to this comparison discussed

only secondary data.18,66,93

31

2.4.1. Migration and ethnic disparities

The protective effect in the immigrant generation has a clear gradient: It is stronger for Black

migrants, still present among Hispanics, but virtually absent among Asians and Whites. This

gradient mirrors the ethnic group hierarchy in the US, which places people of African descent at

the bottom, Hispanics in the middle, and gives (East) Asians a favourable treatment close to that

of Whites. 101,102 These findings are at odds with the classical assimilation theory that predicts a

convergence of the outcomes of migrant groups towards the level observed in the mainstream

White society. 103 Instead, the observed pattern is more consistent with the segmented

assimilation theory that suggests that migrants are selectively incorporated into the system of

stratification of the American society based on their ethnic affiliation. 102

The better birth outcomes of foreign-born Blacks versus their US-born counterparts cannot be

explained by the ‘genetic hypothesis’, which would predict that US-born Blacks be an

intermediate risk group between foreign-born Blacks and US-born Whites because of

intermarriage and genetic mixing over previous generations. 6,83 Among the environmental

explanations, assimilation theories cannot fully account for US-Black disadvantage, since these

theories focus on how migrants and their offspring are incorporated into the host society 102,103

and about 97% of US-born Blacks were fourth or higher generation in 1990. 98 A few studies

have proposed a socio-historic hypothesis, pointing to continuous exposure to socioeconomic

and structural discrimination, 44,86,104 from past historical periods to the urban underclass. Such

explanation is consistent with a substantial sociological literature indicating that racial

segregation concentrates deprivation in Black neighbourhoods by concentrating people who fit

negative racial stereotypes and by restricting the poverty created by economic downturns into a

small number of visible minority neighbourhoods, mainly through discrimination in the housing

32

market. 105,106 Residential racial segregation has been positively associated with infant mortality

among Blacks but negatively among Whites,107 and the Black-White gap in PTB was found to be

higher in hypersegregated areas.108

Because international migration barely contributes to the number of Blacks in the US, the

relative advantage of foreign-born Blacks have little impact on the birth outcomes of Blacks as a

whole. In contrast, migrant women contributed to nearly 60% of births among Hispanics, thus

shaping the birth outcomes of this ethnic group.

2.4.2. Migration and region of origin and destination

Regarding subgroups defined by region of origin and destination, Sub-Saharan African and Latin

American and Caribbean migrants were at higher odds of LBW in Europe but not in the US and

South-Asians were at higher odds in both continents, although their disadvantage was somewhat

attenuated in the US. Part of these differences can be explained by the ethnic composition of the

native-born populations in these analyses, defined by their place of birth but not by their ethnic

groups, and by the patterns of emigration. Thus, US-born compare unfavourably with European-

born partly due to the heavier weight of their ethnic minorities. In the same vein, the Latin

American advantage in the US may be driven by the disproportionate representation of Mexicans

in the US, but not in Europe. Low birthweight rates of Mexicans were among the lowest among

Latin American immigrants. 87 It is believed that Mexicans in the US are protected because of

their residential proximity with co-ethnics, social support systems, and cultural orientation,

75,77,84,109 all of which is facilitated by the spatial contiguity with the home country. The

safeguarding of such protective traits may be more difficult to achieve in transatlantic Europe.

33

The reasons for the higher odds of LBW of Sub-Saharan Africans in Europe compared with

those settling in the US are not clear. Differential migration could not be assessed because, with

one exception, 82 studies did not provide information at the country-level. It is unlikely that the

distribution of reported risk factors accounts for the difference, since the rates of anaemia,

tobacco smoking, marital status, maternal education, and low income were comparable in both

continents. 6,12,60,79,82 Unmeasured factors or a differential effect of the receiving environments

may likely play a role. However, the same receiving environment may affect some migrant

groups favourably and others unfavourably, as suggested in a Swedish study.67

2.4.3. Further research

It remains to be determined whether and to what extent the risk of adverse birth outcomes differs

for particular migrant groups according to their actual destination and whether such effect, if

existent, is due to selective migration or to differential exposures in the receiving environment.

Although we did not find strong evidence that the risk of LBW among particular migrant

subgroups differed according to their geographic destination, the strength of the associations

between particular migrant groups and the native-born population differed by continent (Table

2.4). While our analyses suggest that these differences are mainly driven by differences in the

baseline risk of the European and American populations, the existence of differences in the risk

of adverse birth outcomes within migrant groups according to place of migration remains a

plausible hypothesis. A recently published systematic review 110 reported that the incidence of

adverse birth outcomes among immigrants to European countries were higher in receiving

countries with weak integration policies, such as the U.K. Although this review seems to support

the existence of receiving country-level determinants, confounding is likely to partially account

for its findings, since authors did not control for country of birth. This limitations is of concern

34

when we realized that about 60% of births to immigrants included in the review were born in the

U.K., a country which immigration is dominated by South Asians and Sub-Saharan Africans,89

two groups at high risk of adverse birth outcomes.

Although the comparison between migrants and majority populations may be of interest in itself

for highlighting disparities by migrant status as a single category, the marginal effect estimate of

being foreign-born is actually an average of several groups with high variation in their risk of

birth outcomes. Future research should thus strive to distinguish subgroups defined by their

regions and, when feasible, by their countries of origin since there may be heterogeneity between

countries within the same world region.87,90 Distinguishing subgroups within the receiving-

country population is also recommended, especially in countries highly stratified by

race/ethnicity such as the US.96 Our analyses imply that the definition of the migrant groups and

the choice of the reference groups have a decisive impact on the direction and strength of the

effect estimates for the migrant groups.

Further research on migration and adverse birth outcomes may advance knowledge by examining

why some migrant groups experience poor outcomes and why others do not and what are the

dynamics leading to worse outcomes among the offspring of some migrant groups. Future

studies will benefit from obtaining longitudinal measurements on migrants, including pre-

migration characteristics and circumstances of immigration, and social environment, medical

care and health behaviour after arrival.

35

2.5. References



(2) Bona G, Zaffaroni M, Cataldo F et al. Infants of immigrant parents in Italy. A national multicentre case

control study. Panminerva Medica. 2001;43(3):155-159.





(4) Sanchez BM, Cano FC, Garcia Garcia MC, Yep CG, Perez SE. [Immigration, breastfeeding and smoking

habit]. An Pediatr (Barc ). 2008;68(5):462-465.

(5) Stoltenberg C, Magnus P. Children with low birth weight and low gestational age in Oslo, Norway:

immigration is not the cause of increasing proportions. Journal of Epidemiology & Community Health.

1995;49(6):588-593.

(6) David RJ, Collins JW, Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks,

and U.S.-born whites.[see comment]. New England Journal of Medicine. 1997;337(17):1209-1214.



(8) Wingate MS, Alexander GR. The healthy migrant theory: variations in pregnancy outcomes among US-

born migrants. SOC SCI MED. 2006;62(2):491-498.

(9) Stroup DF, Berlin JA, Morton SC et al. Meta-analysis of observational studies in epidemiology: a proposal

for reporting. Meta-analysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA.

2000;283(15):2008-2012.

36

(10) Harris RP, Helfand M, Woolf SH et al. Current methods of the US Preventive Services Task Force: a

review of the process. Am J Prev Med. 2001;20(3 Suppl):21-35.

(11) Delvaux T, Buekens P, Thoumsin H, Dramaix M, Collette J. Cord C-peptide and insulin-like growth factor-

I, birth weight, and placenta weight among North African and Belgian neonates. American Journal of

Obstetrics & Gynecology. 2003;189(6):1779-1784.

(12) Gayral-Taminh M, Arnaud C, Parant O et al. [Pregnancy and labor of women born in Maghreb and Black

Africa followed to delivery at the Maternity Hospital of Toulouse]. [French]. Journal de Gynecologie,

Obstetrique et Biologie de la Reproduction. 1999;28(5):462-471.

(13) Harding S, Rosato MG, Cruickshank JK. Lack of change in birthweights of infants by generational status

among Indian, Pakistani, Bangladeshi, Black Caribbean, and Black African mothers in a British cohort

study. Intl J Epid. 2004;33(6):1279-1285.

(14) Harding S, Santana P, Cruickshank JK, Boroujerdi M. Birth weights of black African babies of migrant and

nonmigrant mothers compared with those of babies of European mothers in Portugal. Ann Epidemiol.

2006;16(7):572-579.

(15) Margetts BM, Mohd YS, Al Dallal Z et al. Persistence of lower birth weight in second generation South

Asian babies born in the United Kingdom. Journal of Epidemiology & Community Health. 2002;56(9):684-

687.

(16) Moore WM, Bannister RP, Ward BS, Hillier VF, Bamford FN. Fetal and postnatal growth to age 2 years by

mother's country of birth. Early Human Devel. 1995;42(2):111-121.

(17) Rizzo N, Ciardelli V, Gandolfi CG et al. Delivery and immigration: the experience of an Italian hospital.

European Journal of Obstetrics, Gynecology, & Reproductive Biology. 2004;116(2):170-172.

(18) Roville-Sausse F, Truc JB, Jacob D. [Maternal weight gain during pregnancy in various immigrant

communities living in France]. [French]. Revue d Epidemiologie et de Sante Publique. 2001;49(5):439-447.

(19) Vangen S, Stray-Pedersen B, Skrondal A, Magnus P, Stoltenberg C. High risk of cesarean section among

ethnic Filipinos: an effect of the paternal contribution to birthweight? Acta Obstetricia et Gynecologica

Scandinavica. 2003;82(2):192-193.

37

(20) Zambrana RE, Scrimshaw SC, Collins N, Dunkel-Schetter C. Prenatal health behaviors and psychosocial

risk factors in pregnant women of Mexican origin: the role of acculturation. Am J Pub Health.

1997;87(6):1022-1026.

(21) Burton AJ, Lancaster P. Obstetric profiles and perinatal mortality among Pacific Island immigrants in New

South Wales, 1990-93. Australian & New Zealand Journal of Public Health. 1999;23(2):179-184.

(22) Checa MA, Peiro R, Pascual J, Carreras R. Drug intake behaviour of immigrants during pregnancy.

European Journal of Obstetrics, Gynecology, & Reproductive Biology. 2005;121(1):38-45.

(23) Dhawan S. Birth weights of infants of first generation Asian women in Britain compared with second

generation Asian women.[see comment]. BMJ. 1995;311(6997):86-88.

(24) Diani F, Forestieri C, Foschi F, Zaffagnini S, Scollo M, Turinetto A. [Assisted labor among non European

community pregnant women at the Gynecology and Obstetrics Clinic of the Verona University]. [Italian].

Minerva Ginecologica. 2000;52(11):447-457.

(25) Diani F, Zanconato G, Foschi F, Turinetto A, Franchi M. Management of the pregnant immigrant woman

in the decade 1992-2001. Journal of Obstetrics & Gynaecology. 2003;23(6):615-617.

(26) Forna F, Jamieson DJ, Sanders D, Lindsay MK. Pregnancy outcomes in foreign-born and US-born women.

International Journal of Gynaecology & Obstetrics. 2003;83(3):257-265.

(27) Gunton JE, Hitchman R, McElduff A. Effects of ethnicity on glucose tolerance, insulin resistance and beta

cell function in 223 women with an abnormal glucose challenge test during pregnancy. Australian & New

Zealand Journal of Obstetrics & Gynaecology. 2001;41(2):182-186.

(28) Howard DL, Marshall SS, Kaufman JS, Savitz DA. Variations in low birth weight and preterm delivery

among blacks in relation to ancestry and nativity: New York City, 1998-2002. Pediatrics.

2006;118(5):e1399-e1405.

(29) Kelaher M, Jessop DJ. Differences in low-birthweight among documented and undocumented foreign-born

and US-born Latinas. SOC SCI MED. 2002;55(12):2171-2175.

38

(30) Kuvacic I, Skrablin S, Hodzic D, Milkovic G. Possible influence of expatriation on perinatal outcome. Acta

Obstetricia et Gynecologica Scandinavica. 1996;75(4):367-371.

(31) Lalchandani S, MacQuillan K, Sheil O. Obstetric profiles and pregnancy outcomes of immigrant women

with refugee status.[see comment]. Ir Med J. 2001;94(3):79-80.

(32) Leslie JC, Diehl SJ, Galvin SL. A comparison of birth outcomes among US-born and non-US-born

Hispanic Women in North Carolina. Matern Child Health J. 2006;10(1):33-38.

(33) Lopez Gallego MF, Gomez HT, Manzanares GS, Martinez GN, Puertas PA, Montoya VF. Preterm delivery

and immigration. Ciencia Ginecologika. 2005;. 9(4).

(34) Manganaro R, Mami C, Palmara A et al. [Health status of neonates born to immigrants at the University

Polyclinic of Messina in 1993-1998. A case-control study]. [Italian]. Pediatria Medica e Chirurgica.

2000;21(4):197-198.

(35) Martin I, I, Lopez Vilchez MA, Lozano BJ, Mur SA. [Perinatal outcomes in immigrant women]. [Spanish].

Anales de Pediatria. 2006;64(6):550-556.

(36) Medda E, Baglio G, Guasticchi G, Spinelli A. [Reproductive health of immigrant women in the Lazio

region of Italy]. [Italian]. Annali Dell'Istituto Superiore di Sanita. 2002;38(4):357-365.

(37) Miceli M, Di Lallo D. [The reproductive health of immigrant women: a study in the city of Rome in 1982-

1992]. [Italian]. Epidemiologia e Prevenzione. 1996;20(2-3):80-82.

(38) Mitchell J, Mackerras D. The traditional humoral food habits of pregnant Vietnamese-Australian women

and their effect on birth weight. Australian Journal of Public Health. 1995;19(6):629-633.

(39) Nedic B, Loncar S, Ravic J, Komazec J. [Deliveries in the outpatient birthing facility in Ruma during 1989

and between 1992 and 1995]. [Croatian]. Medicinski Pregled. 1999;52(1-2):53-56.

(40) Perez CS, Munoz AN, Robledo SA et al. [Characteristics of immigrant women and their neonates].

[Spanish]. Anales de Pediatria. 2004;60(1):3-8.

39

(41) Porta RP, Lukic A, Nobili F. Ten year survey of the deliveries of immigrant women residing in Rome.

International Journal of Gynaecology & Obstetrics. 1996;52(3):281-282.

(42) Rumbaut R, Weeks J. Unraveling a public health enigma: Why do immigrants experience superior perinatal

health outcomes. Research in the Sociology of Health Care. 1996;13B:337-391.

(43) Sarnoff R, Adams E, Sarnoff R, Adams E. Racial and ethnic disparities in the discordance between

women's assessment of the timing of their prenatal care entry and the first trimester standard. Matern Child

Health J. 2001;5(3):179-187.

(44) Singh GK, Yu SM. Adverse pregnancy outcomes: differences between US- and foreign-born women in

major US racial and ethnic groups. Am J Pub Health. 1996;86(6):837-843.

(45) Sullivan J, Sullivan J. Vietnamese and Australian birth outcomes. Australian Nursing Journal.

1998;5(11):33.

(46) Versi E, Liu KL, Chia P et al. Obstetric outcome of Bangladeshi women in east London. British Journal of

Obstetrics & Gynaecology. 1995;102(8):630-637.

(47) Yoong W, Wagley A, Fong C, Chukwuma C, Nauta M. Obstetric performance of ethnic Kosovo Albanian

asylum seekers in London: a case-control study. Journal of Obstetrics & Gynaecology. 2004;24(5):510-

512.

(48) Yoong W, Kolhe S, Karoshi M, Ullah M, Nauta M. The obstetric performance of United Kingdom asylum

seekers from Somalia: a case-control study and literature review. International Journal of Fertility &

Womens Medicine. 2005;50(4):175-179.

(49) Lauderdale DS. Birth outcomes for Arabic-named women in California before and after September 11.

Demography. 2006;43(1):185-201.

(50) Hessol NA, Fuentes-Afflick E. The perinatal advantage of Mexican-origin Latina women. Annals of

Epidemiology. 2000;10(8):516-523.

40



(52) Fuentes-Afflick E, Hessol NA, Perez-Stable EJ. Testing the epidemiologic paradox of low birth weight in

Latinos. Archives of Pediatrics & Adolescent Medicine. 1999;153(2):147-153.

(53) Berkowitz GS, Blackmore-Prince C, Lapinski RH et al. Risk factors for preterm birth subtypes.

Epidemiology. 1998;9(3):279-285.

(54) Buekens P, Notzon F, Kotelchuck M, Wilcox A. Why do Mexican Americans give birth to few low-birth-

weight infants? Am J Epid. 2000;152(4):347-351.

(55) Korenbrot CC, Dudley RA, Greene JD. Changes in births to foreign-born women after welfare and

immigration policy reforms in California. Matern Child Health J. 2000;4(4):241-250.

(56) Panagopoulos P, Tsoukalos G, Economou A et al. Delivery and immigration: the experience of a Greek

Hospital. Clinical & Experimental Obstetrics & Gynecology. 2005;32(1):55-57.


birthweight among 5 ethnic groups in California.[see comment]. Am J Pub Health. 2001;91(11):1808-1814.

(58) Reime B, Ratner PA, Tomaselli-Reime SN, Kelly A, Schuecking BA, Wenzlaff P. The role of mediating

factors in the association between social deprivation and low birth weight in Germany. SOC SCI MED.

2006;62(7):1731-1744.

(59) Dejin-Karlsson E, Ostergren PO. Country of Origin, Social Support and the Risk of Small for Gestational

Age Birth. Scand J Public Health. 2004;32:442-449.

(60) Essen B, Hanson BS, Ostergren PO, Lindquist PG, Gudmundsson S. Increased perinatal mortality among

sub-Saharan immigrants in a city-population in Sweden. Acta Obstetricia et Gynecologica Scandinavica.

2000;79(9):737-743.

(61) Foix-L'Helias L, Blondel B. Changes in risk factors of preterm delivery in France between 1981 and 1995.

Paediatr Perinat Epidemiol. 2000;14(4):314-323.

41

(62) Zeitlin J, Bucourt M, Rivera L, Topuz B, Papiernik E. Preterm birth and maternal country of birth in a

French district with a multiethnic population. BJOG: an International Journal of Obstetrics &

Gynaecology. 2004;111(8):849-855.

(63) National Center for Health Statistics. Technical Appendix. Vital Statistics of the United States, 2002,

natality. 2003. Hyattsville, Maryland.

(64) United Nations. United Nations World Macro Regions and Components. [http://www un

org/depts/dhl/maplib/worldregions htm]. 2000. Available at:

http://www.un.org/depts/dhl/maplib/worldregions.htm.

(65) United Nations Children's Fund and World Health Organization. Low Birthweight: Country, Regional and

Global Estimates. 2004. New York, UNICEF.

(66) Guendelman S, Buekens P, Blondel B, Kaminski M, Notzon FC, Masuy-Stroobant G. Birth outcomes of

immigrant women in the United States, France, and Belgium. Matern Child Health J. 1999;3(4):177-187.




(68) Hox J. Multilevel Analysis. Techniques and Applications. Mahwah, NJ: Lawrence Erlbaum Associates;

2002.

(69) Raudenbush SW, Bryk AS. Hierarchical Linear Models. Applications and Data Analysis Methods. second

ed. Thousand Oaks, CA: Sage Publications; 2002.

(70) Thompson SG, Turner RM, Warn DE. Multilevel models for meta-analysis, and their application to

absolute risk differences. Stat Methods Med Res. 2001;10(6):375-392.

(71) Higgins JP, Thompson SG. Controlling the risk of spurious findings from meta-regression. Stat Med.

2004;23(11):1663-1682.

42

(72) Thompson SG, Higgins JP. How should meta-regression analyses be undertaken and interpreted? Stat Med.

2002;21(11):1559-1573.

(73) Snijders T, Bosker RJ. Multilevel Analysis. An Introduction to basic and advanced multilevel modeling.

London: Sage Publications; 1999.

(74) Alexander GR, Mor JM, Kogan MD, Leland NL, Kieffer E. Pregnancy outcomes of US-born and foreign-

born Japanese Americans. Am J Pub Health. 1996;86(6):820-824.



109.

(76) Cocroft JR, Hauck WW, Cosler L et al. The effect of ethnicity and maternal birthplace on small-for-

gestational-age deliveries to HIV-infected women. Journal of Urban Health. 2002;79(1):147-160.







(80) Fuentes-Afflick E, Hessol NA, Perez-Stable EJ. Maternal birthplace, ethnicity, and low birth weight in

California. Archives of Pediatrics & Adolescent Medicine. 1998;152(11):1105-1112.

(81) Gould JB, Madan A, Qin C, Chavez G. Perinatal outcomes in two dissimilar immigrant populations in the

United States: a dual epidemiologic paradox. Pediatrics. 2003;111(6 Pt 1):e676-e682.

(82) Johnson EB, Reed SD, Hitti J, Batra M. Increased risk of adverse pregnancy outcome among Somali

immigrants in Washington state. American Journal of Obstetrics & Gynecology. 2005;193(2):475-482.

43

(83) Kramer MS, Ananth CV, Platt RW, Joseph KS. US Black vs White disparities in foetal growth:

physiological or pathological? Int J Epidemiol. 2006;35(5):1187-1195.

(84) Landale NS, Oropesa, Gorman BK. Immigration and infant health:birth outcomes of immigrant and native-

born women. Children of Immigrants: Health, Adjustment and Public Assistance. Washington, DC:

National Academy Press, 1999: 244-85.

(85) Madan A, Palaniappan L, Urizar G, Wang Y, Fortmann SP, Gould JB. Sociocultural factors that affect

pregnancy outcomes in two dissimilar immigrant groups in the United States. Journal of Pediatrics.

2006;148(3):341-346.

(86) Pallotto EK, Collins JW, Jr., David RJ. Enigma of maternal race and infant birth weight: a population-

based study of US-born Black and Caribbean-born Black women. Am J Epid. 2000;151(11):1080-1085.

(87) Rosenberg TJ, Raggio TP, Chiasson MA. A further examination of the "epidemiologic paradox": birth

outcomes among Latinas. Journal of the National Medical Association. 2005;97(4):550-556.

(88) Buekens P, Masuy-Stroobant G, Delvaux T. High birthweights among infants of north African immigrants

in Belgium. Am J Pub Health. 1998;88(5):808-811.

(89) Collingwood BA. Trends in live births by mother's country of birth and other factors affecting low

birthweight in England and Wales, 1983-2001. Health Statistics Quarterly. 2004;(23):25-33.

(90) Fuentes-Afflick E, Hessol NA. Impact of Asian ethnicity and national origin on infant birth weight. Am J

Epidemiol. 1997;145(2):148-155.

(91) Gissler M, Pakkanen M, Olausson PO. Fertility and perinatal health among Finnish immigrants in Sweden.

SOC SCI MED. 2003;. 57(8).

(92) Vangen S, Stoltenberg C, Skjaerven R, Magnus P, Harris JR, Stray-Pedersen B. The heavier the better?

Birthweight and perinatal mortality in different ethnic groups. Intl J Epid. 2002;31(3):654-660.

(93) Vahratian A, Buekens P, Delvaux T et al. Birthweight differences among infants of North African

immigrants and Belgians in Belgium. European Journal of Public Health. 2004;14(4):381-383.

44

(94) Baumeister L, Marchi K, Pearl M, Williams R, Braveman P. The validity of information on "race" and

"Hispanic ethnicity" in California birth certificate data. Health Serv Res. 2000;35(4):869-883.

(95) Schoendorf KC, Parker JD, Batkhan LZ, Kiely JL. Comparability of the birth certificate and 1988 Maternal

and Infant Health Survey. Vital Health Stat 2. 1993;(116):1-19.

(96) Kaufman JS, Cooper RS. Commentary: considerations for use of racial/ethnic classification in etiologic

research. Am J Epidemiol. 2001;154(4):291-298.

(97) Senior PA, Bhopal R. Ethnicity as a variable in epidemiological research. BMJ. 1994;309(6950):327-330.

(98) Edmonston B, Passel JS. Immigration and immigrant generations in population projections. Int J Forecast.

1992;8(3):459-476.

(99) Kramer MS. The epidemiology of adverse pregnancy outcomes: an overview. J Nutr. 2003;133(5 Suppl

2):1592S-1596S.

(100) Rubalcava LN, Teruel GM, Thomas D, Goldman N. The healthy migrant effect: new findings from the

Mexican Family Life Survey. Am J Public Health. 2008;98(1):78-84.

(101) Massey DS, Denton NA. Trends in the Residential Segregation of Blacks, Hispanics, and Asians: 1970-

1980. American Sociological Review. 1987;52:802-825.

(102) Zhou M. Segmented Assimilation: Issues, Controversies, and Recent Research on the New Second

Generation. Int Migr Rev. 1997;31(4):975-1008.

(103) Rumbaut RG. Assimilation and its discontents: between rhetoric and reality. Int Migr Rev. 1997;31(4):923-

960.

(104) James SA. Racial and ethnic differences in infant mortality and low birth weight. A psychosocial critique.

Ann Epidemiol. 1993;3(2):130-136.

(105) Massey DS. American Apartheid: Segregation and the Making of the Underclass. The American Journal of

Sociology. 1990;96(2):329-357.

45

(106) Massey DS, Gross AB, Shibuya K. Migration, Segregation, and the Geographic Concentration of Poverty.

American Sociological Review. 1994;59(3):425-445.

(107) LaVeist TA. Linking Residential Segregation and Infant Mortality in U.S. Cities. Sociology and Social

Research. 1989;73:90-94.

(108) Osypuk TL, cevedo-Garcia D. Are racial disparities in preterm birth larger in hypersegregated areas? Am J

Epidemiol. 2008;167(11):1295-1304.

(109) Peak C, Weeks JR. Does community context influence reproductive outcomes of Mexican origin women in

San Diego, California? J Immigr Health. 2002;4(3):125-136.

(110) Bollini P, Pampallona S, Wanner P, Kupelnick B. Pregnancy outcome of migrant women and integration

policy: A systematic review of the international literature. Soc Sci Med. 2009;68(3):452-461.

46

Chapter 3 Increase in Preterm Births Associated with Duration of Residence

among Immigrants living in Ontario Metropolitan Areas

Abstract

The proportion of live births to migrant women in industrialized countries has been growing

during the last decades. Yet, the determinants of adverse birth outcomes among immigrants are

not well understood. The ‘convergence hypothesis’ suggests that health outcomes of immigrants

will approach the level observed in the native population with increasing time in the new

environment. We linked birth data (2002-2007) with an immigration database of all legal

immigrants to Ontario, Canada, who obtained their permanent residence between 1985 and 2000

to examine the association between duration of residence and birth outcomes among immigrants

(N=83,233) and compare it with the birth outcomes of non-immigrants (N=314,237). We used

hierarchical models to account for the clustering of births into maternal countries of birth. After

adjustment for pregnancy and immigration characteristics, duration of residence was associated

with increases in low birth weight [5-year adjusted odds ratio (95% CI)]: 1.08 (1.03-1.13),

preterm birth: 1.14 (1.10-1.19), and small preterm: 1.15 (1.09-1.21) but not with small for

gestational age: 0.99 (0.96-1.02). Our findings suggest that the deterioration of the birth

outcomes was driven by a shortening of gestational age with time spent in Canada.

47

3.1. Introduction

Women constitute half – almost 95 million - of all international migrants.1 The percentage of

births among migrant women in most industrialized countries has shown upward trends in the

last two decades.2-4 Currently, immigrant women contribute to more than one fifth of all live

births in the United States3 and several European countries.5

Much of the research on the effects of acculturation on birth outcomes has been carried out in the

US, particularly among women of Mexican descent, who experienced deterioration in their birth

outcomes with increased acculturation.6,7 Acculturation has been measured by acculturation

scales7, and proxies such as language use8, nativity8-11 (i.e., foreign-born versus native-born or

country of birth), and duration of residence.12-14 Language use was identified as the main

component of acculturation6 and it was suggested that it may modify the effect of nativity.8

Differences in birth outcomes between foreign-born and native-born (second generation) women

of the same ethnicity may reflect selection bias (the healthy migrant effect) or the effect of the

receiving country environment, either via acculturation or environmental exposures.

While nativity is useful to measure changes in the outcomes from one generation to the next, it

cannot assess changes in the outcomes within first generation migrants after arrival.

Comparisons by nativity status may be confounded by selection of healthy migrants, cohort

effects, or by duration of residence of foreign-born women if there is an association between

duration of residence and birth outcomes.

Data needed to assess duration of residence are scarce, thus limiting the number of studies on

this issue. Duration of residence was associated with a deterioration of birth outcomes among

foreign-born Mexicans in Washington12 and California.13 However, it is not clear whether

48

women from other regions of the world are equally affected by the length of residence in the

receiving countries. Finns lowered their risk of low birthweight and preterm birth after 3 years of

stay in Sweden but Sub-Saharan Africans experienced the opposite, suggesting that this relation

may vary by the maternal world region of origin.14 In addition, it has been suggested that the

relation between acculturation and low birthweight may not be linear.8,15 The few studies

assessing the association between duration of residence in the host country and low birthweight

and preterm birth have measured duration of residence as a dichotomous variable, with recent

immigrants defined by less than two,12 three,14 and five13 years versus the remaining immigrants

(long-term residents). The length of the observation period may impact on the study conclusions.

It is unknown whether duration of residence considered in all its length has a threshold effect, a

dose-response effect, or a non-linear effect on birth outcomes of immigrants, and whether such

relations vary according to the maternal world region of origin.

One potential pattern is provided by the ‘convergence hypothesis’, inspired on the studies on

coronary heart disease and stroke gradients of Japanese men living in Japan, Hawaii, and

California.16 The convergence hypothesis suggests that health outcomes of immigrants will tend

to converge over time towards the level observed in the host population, presumably via changes

in health-related behaviours. While this hypothesis seems to hold for overweight/obesity,17,18 and

behavioural risk factors18-20 the evidence is not consistent for mortality21, and it is not known

whether it applies to birth outcomes.

Our objectives were to examine the relation between immigrants’ duration of residence in Urban

Ontario and adverse birth outcomes (low birthweight and preterm birth and its components, and

small for gestational age) across 20 years since arrival, overall and by maternal world regions of

origin, and to make comparisons with the non-immigrant population. Urban Ontario is an

49

appropriate setting to study immigrants’ health as Ontario is Canada’s most populated province

and receives annually about half of all immigrants to Canada, with more than 90 % of them

concentrated in urban areas.22

3.2. Methods

3.2.1. Data sources

The Discharge Abstract Database (DAD) of the Canadian Institute for Health Information (CIHI)

compiles information on admissions/services/discharges of all acute Ontario hospitals, where

most deliveries take place, excepting home births.

Birth and maternal records were internally merged according to an algorithm described

elsewhere,23 updated to reflect the changes from ICD-9 to ICD-10-CA, resulting in 96% of

mothers with a valid match to an infant in the fiscal years 2002/2003-2006/2007. Although the

DAD includes a deterministic linkage between the records of the mother and the child since

fiscal 2002/2003 the resulting match was in between 80% and 90%, and therefore ICES

programmers preferred the probabilistic linkage as the most sensitive. Further details on the data

sources are in Appendix 3.A.

We extracted 474,614 singleton live births (see Appendix 3.B. for details about the exclusion of

stillbirths and multiple births) born to mothers living in any of the 11 Ontario Census

Metropolitan Areas (Great Sudbury, Hamilton, Kingston, Kitchener, London, Oshawa, St.

Catherines-Niagara, Ottawa-Gatineau, Thunder Bay, Toronto, and Windsor)24 at the time of

delivery, between April 1, 2002 and March 31, 2007.

50

Data on immigration were contained in the Landed Immigrant Data System (LIDS), which is the

official immigration registry compiled by Citizenship and Immigration Canada (CIC) and

contains information on sociodemographic and immigration characteristics. These records were

linked with the registry of the Ontario Health Insurance Plan (OHIP), which provides universal

access to nearly all physician and hospital services (except for asylum seekers and during the

first 3 months’ residence). This linkage matched 84% of individuals whose intended destination

was Ontario and that obtained their legal permanent residence from January 1, 1985 to December

31, 2000. Many unmatched individuals may have moved out of the province shortly upon arrival.

In order to avoid misclassification of immigrant status regarding immigrants obtaining their

permanent residence after December 2000, we excluded 74,961 infants whose mothers were first

registered into the Ontario Health Insurance Plan after March 31, 2001 (to account for the 3-

month waiting period), who may be newcomers either from abroad or from other provinces. We

further excluded births weighing less than 500 grams and more than 6000 grams because of the

high likelihood of data errors (N=360), with missing information on the outcomes or gestational

age (N=125), with gestational age less than 22 and more than 43 completed weeks (N=72), with

missing information on infant sex, parity, or maternal age (N=54), on maternal

sociodemographic and immigration characteristics (N=576), and immigrants classified as ‘other’

(N=487). These data were finally merged with small-area data (census tracts) from the 2001

Canadian census. After excluding 509 records to which census information could not be

assigned, our study population for analyses comprised 397,470 singleton live births (21% of

them born to immigrant mothers). A flowchart of the data exclusions is shown in Appendix 3.C.

51

3.2.2. Outcomes

We used several adverse birth outcomes, all defined as categorical. Low birthweight (LBW) was

defined as a birth with less than 2,500 grams, very low birthweight (VLBW) with less than 1,500

grams, and moderately low birthweight (MLBW) within the range 1,500-2,499 grams. Preterm

birth (PTB) was defined as a delivery before 37 completed weeks of gestation, very preterm birth

(VPTB) before 32 weeks, and moderately preterm birth (MPTB) as 32 weeks or more but before

37 weeks. We included small preterm birth (SPTB) defined as infants having both LBW and

PTB, because these babies are at higher risk of subsequent adverse outcomes, such as infant

mortality.25 We defined small for gestational age (SGA) as a birth weight lower than the 10th

lowest percentile of the most recent Canadian sex- and gestational age-specific birthweight

distribution,26 because this measure represents a proxy for fetal growth that is relatively

independent of gestational age.

3.2.3. Predictors

We assessed the amount of exposure to a Canadian setting by duration of residence in Canada,

defined as the time (in days) since arrival to delivery. We also modeled duration in completed

years and in approximately 5-year duration groups (15 months to 4 years, 5 to 9 years, 10 to 14

years, and 15 years and more).

As potential confounders, we considered from hospital records infant gender (male as referent),

maternal age at delivery (15-19, 20-24, 25-29, 30-34 [referent], 35-39, and ≥ 40 years), parity

(primiparae versus multiparae). We conceptualized maternal morbidity during pregnancy as a

potential mediator of the effect of duration of residence on preterm birth by including genito-

52

urinary infection (ICD-10: O230-O235, O239), pregnancy-induced hypertension (O13, O140,

O141, O149), incompetent cervix (O343), and placental abruption (O450, O458, O459).

Immigrant characteristics that may confound the association between duration of residence and

adverse birth outcomes included maternal country of birth, maternal age at arrival (<12, 12-18,

19-25, 26-30, and > 30 years), high school graduation (yes as referent), and marital status

(married or common law [referent] versus single, widowed, or separated), immigrant class

(economic class, family class [referent], and refugee status), and knowledge of any official

Canadian language (English or French [referent] versus none). These characteristics were

measured at landing and based on legal documentation provided by the immigrants during the

application process, with the only exception of language knowledge, which was self-reported.

Countries of birth were grouped into world regions following the UNICEF classification.27 We

also assigned each mother a material deprivation score based on their place of residence at

delivery. The material deprivation score was constructed by summarizing information on six

census variables28 (percentage of population aged ≥ 20 years without high school graduation,

percentage of lone parent families, percentage of families receiving government transfer

payments, percentage of population aged ≥ 15 years unemployed, percentage living below the

low income cut off, and percentage of homes needing major repair) at the census-tract level by

means of factor analysis. The resulting continuous score was standardized thus having a mean of

0 and a standard deviation of 1. We collapsed it into tertiles of approximately equal number of

census tracts for easier interpretation. Census tracts (our neighbourhoods) were relatively stable

urban neighbourhoods with a typical population of 2500 to 8000 and were relatively

homogeneous in population characteristics and living conditions.

53


We inspected the shape of the relation between duration of residence (measured as days from

arrival to delivery) and all birth outcomes using bivariate plots and logistic models with linear,

quadratic, and cubic terms, as well as with dummy variables (approximately 5-year duration

groups). A linear specification of duration resulted in a better fit assessed by the likelihood ratio

test.

Selection of variables for covariate adjustment for the association between duration of residence

and preterm birth among immigrants was based on directed acyclic graphs (DAGs).29 We

checked the hypothesized relation between the variables in our conceptual model by empirically

assessing the association between the variables before conducting the backdoor test for

sufficiency.29,30 All potential confounders were included in the adjusted models, with the

exception of neighbourhood material deprivation because it was a mediator and infant sex

because it was not directly or indirectly associated with the exposure. We did not consider health

behaviours and maternal morbidity during pregnancy because they are considered to mediate the

effects of duration on birth outcomes. Causal diagrams are shown in Appendix 3.D.

We conducted multilevel logistic regression analyses using GLIMMIX in SAS 9.1 (SAS

Institute, Cary, NC) to account for the clustering of births within maternal countries of birth

among immigrants, a clustering that was found to be more relevant for immigrants than

clustering into neighbourhoods (see Chapter 4). The use of hierarchical models avoids incurring

in the ecologic fallacy 31 that consists in drawing inferences at the individual level (level 1) using

data measured at higher levels (level 2) (e.g., countries, neighbourhoods) and allows the

estimation of cross-level effects (e.g., country of origin effects on birth outcomes). Hierarchical

54

models are useful even when the independent variable of interest is measured at the individual-

level (e.g., duration of residence) because the effect estimate of the level 1 variable on the

outcome is adjusted for the statistical dependency of the individual observations within level 2

units (e.g., maternal countries of origin).32 Failure to take into account the dependence of

individual observations within higher-level units may lead to underestimate standard errors and

misleading conclusions. In fact, ordinary least squares (OLS) regression is one special case of the

multilevel model in which the level-2 variance equals zero, meaning that all the variability is

inter-individual and there is no inter-group variability.

3.3. Results

Immigrants exhibited higher proportions of adverse outcomes with increasing duration of

residence, with the exception of SGA (Table 3.1). Immigrants also differed in marital status,

education, and knowledge of official languages, according to duration of residence, but part of

these differences were due to the fact that these variables were measured at arrival and therefore

affected by the maternal age at arrival. Yet, the trends across duration groups remained

significant after restricting the population to women aged 20 years or more at arrival (not

shown). Higher proportions of refugees and immigrants from Latin America and industrialized

countries with more years of residence reflect changing immigration patterns over time, more

recently dominated by immigrants from East and South Asia.

55

Table 3.1: Characteristics of the study population, by migrant status, and duration of residence, Urban

Ontario, live births 2002-2007

Immigrants by duration of residence

Non-immigrants

All Immigrants

p-value* < 5 y 5-9 y 10-14 y 15 y + p-trend**

Singleton live births 314,237 83,233

14,555 32,539 23,827 12,312

Outcomes Low birthweight % 4.29 5.34 <.0001 4.61 4.98 5.93 6.03 <.0001 Very low birthweight % 0.63 0.88 <.0001 0.61 0.80 0.99 1.19 <.0001 Moderately low birthweight % 3.66 4.46 <.0001 4.00 4.18 4.94 4.83 <.0001 Preterm birth % 6.25 5.99 0.0059 4.72 5.57 6.61 7.43 <.0001 Very preterm birth % 0.77 0.99 <.0001 0.66 0.91 1.12 1.34 <.0001 Moderately preterm birth % 5.48 5.00 <.0001 4.06 4.66 5.48 6.09 <.0001 Small preterm birth % 2.88 3.24 <.0001 2.39 2.96 3.73 4.03 <.0001 Small for gestational age % 7.68 11.06 <.0001 11.37 11.00 11.01 10.95 0.3145 Infant and maternal characteristics

Infant sex (male) 51.18 51.66 0.0149 51.67 51.78 51.34 51.93 0.9409 Maternal age at delivery < 20 years 3.67 1.83 <.0001 1.06 1.63 2.49 2.00 <.0001 20-24 years 11.42 10.85 <.0001 12.87 9.67 10.77 11.76 0.2868 25-29 years 26.22 27.93 <.0001 33.47 28.73 23.92 27.01 <.0001 30-34 years 36.54 34.01 <.0001 35.67 36.71 31.70 29.37 <.0001 35-39 years 18.53 20.49 <.0001 14.68 19.66 24.49 21.79 <.0001 ≥ 40 years 3.62 4.89 <.0001 2.25 3.60 6.63 8.07 <.0001 Primipara 46.54 36.55 <.0001 40.97 33.74 35.79 40.21 0.5541 Neighbourhood material deprivation

Low 44.41 29.66 <.0001 23.84 30.01 29.62 35.97 <.0001 Med 29.18 28.44 <.0001 30.14 28.72 27.57 27.39 <.0001 High 26.41 41.90 <.0001 46.02 41.27 42.81 36.94 <.0001 Maternal morbidity Genito-urinary infection 1.95 1.55 <.0001 2.08 1.48 1.50 1.20 <.0001 Pregnancy-induced hypertension 0.97 0.77 <.0001 0.51 0.69 0.91 1.04 <.0001

Incompetent cervix 0.31 0.49 <.0001 0.27 0.45 0.59 0.67 <.0001 Abruptio placenta 1.02 0.92 0.0122 0.90 0.90 0.97 0.93 0.5727 Maternal age at arrival < 12 y N/A 9.47 0.03 1.66 11.77 36.79 <.0001 12-18 y 22.28 4.79 16.22 32.76 38.70 <.0001 19-25 y 40.59 39.53 46.57 42.44 22.51 <.0001 26-30 y 20.92 36.47 27.83 11.78 1.96 <.0001 30 y + 6.74 19.18 7.72 1.25 0.04 <.0001 Unmarried / not cohabiting at arrival N/A 56.06 18.35 45.59 76.17 89.38 <.0001

No High School graduation at arrival N/A 63.36 38.19 54.66 77.53 88.69 <.0001

Knowledge of English/French at arrival N/A 56.60 51.90 55.63 58.53 61.03 <.0001

Immigrant Class N/A Economic 28.73 38.72 27.01 23.01 32.50 <.0001 Family 57.42 57.32 63.99 56.94 41.11 <.0001 Refugee 13.85 3.96 9.00 20.06 26.39 <.0001 World Region of birth N/A Central / East Europe 7.42 9.28 9.64 5.65 2.75 <.0001

56

Immigrants by duration of residence

Non-immigrants

All Immigrants

p-value* < 5 y 5-9 y 10-14 y 15 y + p-trend**

Latin America & Caribbean 18.24 7.18 12.82 23.51 35.45 <.0001 Middle East / North Africa 7.42 9.21 8.23 6.33 5.26 <.0001 East Asia / Pacific 18.32 22.51 19.31 17.06 13.18 <.0001 South Asia 27.05 40.16 32.36 19.91 11.35 <.0001 Sub Saharan Africa 7.70 4.19 7.60 10.56 6.59 <.0001 Industrialized Countries 13.85 7.47 10.64 16.98 25.41 <.0001

* Chi square between immigrants versus non-immigrants ** Two-sided p-value of Cochran-Armitage test for trends for binomial proportions across duration groups (when the variable had more than 2 categories, each category was compared to the rest)

Table 3.2. Unadjusted odds ratios (and 95% confidence intervals) for adverse birth outcomes between

non-immigrants and immigrants, overall and by duration of residence in Canada, 2002/2003 to 2006/2007

All immigrants versus Non-immigrants

Immigrants by duration of residence versus non-immigrants

Outcome < 5 years 5 – 9 years 10 -14 years ≥ 15 years OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI) OR (95% CI)

Low birthweight 1.26 (1.22-1.30) 1.08 (1.00-1.16) 1.17 (1.11-1.23) 1.41 (1.33-1.49) 1.43 (1.33-1.55)

Very low birthweight 1.39 (1.28-1.52) 0.97 (0.78-1.20) 1.26 (1.11-1.44) 1.57 (1.37-1.80) 1.90 (1.61-2.25)

Moderately low birthweight 1.23 (1.19-1.28) 1.10 (1.01-1.19) 1.15 (1.09-1.22) 1.37 (1.29-1.46) 1.34 (1.23-1.46)

Preterm birth 0.96 (0.93-0.99) 0.74 (0.69-0.80) 0.88 (0.84-0.93) 1.06 (1.01-1.12) 1.21 (1.12-1.29)

Very preterm birth 1.28 (1.19-1.39) 0.85 (0.69-1.05) 1.17 (1.04-1.33) 1.46 (1.29-1.66) 1.74 (1.49-2.04)

Moderately preterm birth 0.91 (0.88-0.94) 0.73 (0.67-0.79) 0.84 (0.80-0.89) 1.00 (0.95-1.06) 1.12 (1.04-1.21)

Small preterm birth 1.13 (1.08-1.18) 0.82 (0.74-0.92) 1.03 (0.96-1.10) 1.31 (1.22-1.40) 1.42 (1.29-1.55)

Small for gestational age 1.50 (1.46-1.53) 1.54 (1.46-1.63) 1.49 (1.43-1.54) 1.49 (1.43-1.55) 1.48 (1.39-1.57)

Table 3.2 compares the unadjusted odds of adverse birth outcomes of immigrant women (as a

single category and by duration of residence) relative to non-immigrants. Adjustment for infant

sex, maternal age, and parity did not alter the estimates substantially.

57

Comparisons of all immigrants as a single category with non-immigrants represent average

estimates that mask the importance of duration of residence as a determinant of adverse birth

outcomes (except SGA) among immigrants. Immigrant women with short duration of residence

had lower odds of adverse birth outcomes than the average while those with long duration had

higher odds than the average (Table 3.2). In addition, recent immigrant women (< 5 years)

exhibited similar or lower odds of adverse birth outcomes than their Canadian counterparts, but

these advantages were lost after 5 to 14 years, depending on the outcome, and exhibited

consistently higher odds than the reference group after 15 years of residence.

Table 3.3 shows the results of the multilevel models fitted to better evaluate whether the

association between immigrants’ duration of residence and adverse birth outcomes was not due

to confounding by maternal and immigration characteristics. With the exception of small for

gestational age, duration of residence was independently associated with all remaining outcomes.

The absence of an association between duration and SGA when the remaining outcomes were all

associated with duration is suggestive of the presence of some mechanism related to a shortening

of gestational age. To explore this possibility we fitted models for LBW and its components,

further adjusted for preterm birth or gestational age. Results shown in Table 3.4 indicate that the

trends in these outcomes with increasing duration were mainly explained by shorter gestations.

58

Table 3.3: Adjusted odds ratios (and 95% confidence intervals) for adverse birth outcomes by immigrants’

duration of residence in Canada, urban Ontario live births 2002-2007

5-year AOR* Immigrants by duration of residence

outcome <5 years 5 – 9 years 10 -14 years ≥ 15 years p-trend

Low birthweight 1.08 (1.03-1.13) 1.00 1.06 (0.96-1.17) 1.20 (1.07-1.34) 1.17 (1.02-1.33) 0.0044

Very low

birthweight 1.17 (1.06-1.30) 1.00 1.22 (0.95-1.57) 1.34 (1.01-1.78) 1.54 (1.11-2.12) 0.0099

Moderately low

birthweight 1.06 (1.01-1.11) 1.00 1.03 (0.93-1.15) 1.17 (1.04-1.32) 1.10 (0.95-1.27) 0.0527

Preterm birth 1.14 (1.10-1.19) 1.00 1.14 (1.04-1.26) 1.27 (1.14-1.42) 1.39 (1.23-1.58) <.0001

Very preterm birth 1.18 (1.07-1.30) 1.00 1.28 (1.01-1.63) 1.40 (1.07-1.84) 1.60 (1.18-2.17) 0.0037

Moderately

preterm birth 1.13 (1.09-1.19) 1.00 1.12 (1.01-1.24) 1.25 (1.11-1.40) 1.35 (1.18-1.55) <.0001

Small preterm birth 1.15 (1.09-1.21) 1.00 1.18 (1.04-1.34) 1.36 (1.17-1.57) 1.40 (1.18-1.67) <.0001

Small for

gestational age 0.99 (0.96-1.02) 1.00 0.98 (0.92-1.05) 0.97 (0.90-1.05) 0.95 (0.86-1.04) 0.2984

* AOR= adjusted odds ratios, adjusted for maternal age at delivery, parity, immigrant class, country and region of birth, language

knowledge, high school graduation, and unmarried status.

Table 3.4. Odds ratios adjusted for all covariates in Table 3.3 and also adjusted for preterm birth

Immigrants by duration of residence Outcome 5-year AOR*

<5 years 5 – 9 years 10 -14 years ≥ 15 years p-trend

Low birthweight 1.00 (0.95-1.05) 1.00 0.97 (0.87-1.09) 1.06 (0.92-1.21) 0.96 (0.82-1.12) 0.9963

Very low

birthweight 1.05 (0.94-1.17) 1.00 1.08 (0.82-1.41) 1.05 (0.77-1.42) 1.14 (0.81-1.61) 0.5567

Moderately low

birthweight 0.99 (0.94-1.04) 1.00 0.96 (0.85-1.08) 1.05 (0.92-1.20) 0.92 (0.79-1.08) 0.7234

* AOR= adjusted odds ratios, adjusted for maternal age at delivery, parity, immigrant class, country and region of birth, language

knowledge, high school graduation, unmarried status, and preterm birth.

Figure 3.1 depicts predicted probabilities based on a preterm birth 2-level model, with births

nested within maternal countries of birth, including duration of residence as a continuous

variable and adjusted for the variables selected based on the DAG. The flat line represents the

59

crude average preterm birth percent for non-immigrants during the study period, that is, the level

achieved by the host population, against which to compare the level of preterm birth achieved by

the immigrants according to their years of residence in Canada.

Figure 3.1. Predicted probabilities (and 95% CI) of Preterm Birth (2002-2007) among Ontario immigrants,

by duration of residence*

* adjusted for maternal age at delivery, parity, immigrant class, country and region of birth, language knowledge, high school

graduation, unmarried status.

60

Figure 3.2. Gestational age distributions (and means) by duration of residence groups

The frequency distribution of gestational age shifts uniformly to the left (and also their means –

vertical lines) with increasing duration of residence.

These estimates of Table 3.5 were obtained by stratified analyses by world region. Multilevel

modeling was used for all immigrants but not in stratified analysis by world regions due to the

small number of countries within them. For those outcomes previously associated with duration

of residence there were no great differences between world regions. With the exception of SGA,

the direction of the association with duration did not substantially vary across world regions,

being positive or null but not negative.

61

Table 3.5: Adjusted* odds ratios per 5-year increase in duration of residence in Canada (birth to

immigrants 2002-2007), by world region LBW VLBW MLBW PTB VPTB MPTB Small PTB SGA

All immigrants 1.08

(1.03-1.13)

1.17

(1.06-1.30)

1.06

(1.01-1.11)

1.14

(1.10-1.19)

1.18

(1.07-1.30)

1.13

(1.09-1.19)

1.15

(1.09-1.21)

0.99

(0.96-1.02)

By world region

Central/East

Europe

1.26 (1.02-1.57)

1.11

(0.64-1.93)

1.29 (1.02-1.64)

1.19

(0.99-1.42)

1.02

(0.64-1.61)

1.22 (1.00-1.48)

1.32 (1.02-1.71)

1.06

(0.90-1.25)

Latin America &

Caribbean

1.00

(0.92-1.08) 1.11

(0.94-1.31)

0.97

(0.89-1.06)

1.07 (1.00-1.16)

1.10

(0.94-1.29)

1.06

(0.98-1.16)

1.12 (1.01-1.23)

0.94

(0.88-1.00)

Middle East /

North Africa

0.98

(0.81-1.18)

0.79

(0.49-1.28)

1.02

(0.83-1.25)

1.07

(0.90-1.26)

1.26

(0.83-1.89)

1.04

(0.87-1.24)

1.07

(0.85-1.36)

0.83 (0.72-0.94)

East Asia /

Pacific

1.25 (1.12-1.40)

1.10

(0.81-1.50)

1.27 (1.13-1.43)

1.25 (1.13-1.38)

1.05

(0.80-1.40)

1.27 (1.14-1.42)

1.20 (1.04-1.38)

1.08

(1.00-1.18)

South Asia 1.06

(0.97-1.16)

1.45 (1.17-1.79)

1.00

(0.91-1.10)

1.13 (1.03-1.23)

1.32 (1.07-1.63)

1.09

(0.99-1.20)

1.19 (1.06-1.34)

0.96

(0.90-1.02)

Sub Saharan

Africa

1.44 (1.22-1.69)

1.32

(0.94-1.85)

1.45 (1.21-1.75)

1.31 (1.11-1.53)

1.52 (1.08-2.14)

1.24 (1.04-1.48)

1.31 (1.07-1.61)

1.18 (1.04-1.33)

Industrialized

Countries

1.10

(0.97-1.25)

1.28

(0.93-1.77)

1.07

(0.93-1.23)

1.20

(1.07-1.34)

1.24

(0.93-1.64)

1.19

(1.05-1.34)

1.12

(0.95-1.31)

1.08

(0.98-1.17)

* adjusted for maternal age at delivery, parity, immigrant class, language knowledge, high school graduation, and unmarried status.

3.3.1. Sensitivity analyses

In order to evaluate the possibility that the observed associations between duration of residence

and adverse birth outcomes were confounded by cohort effects, we selected a group of

immigrants (N=69,522) that obtained their permanent residence from January 1985 to December

31, 1998, and observed their LBW and PTB over time and the results were consistent with our

main analyses (see Appendix 3.E). In this ‘cohort approach’ we could not adjust for parity

because this variable was not available prior to fiscal 2002/2003 but we adjusted for year of birth

to control for the secular trends in preterm birth over time.33,34

62

Figure 3.3. Predicted probabilities* of preterm birth according to duration of residence, by cohort of arrival

* Based on a stratified multilevel model (country of birth as random intercept) by 4-year cohort of arrival (1985 to 1988, 1989 to

1992, 1993 to 1996, and 1997 to 2000), adjusted for maternal age at delivery, immigrant class, language knowledge, high school

graduation, unmarried status, and year of birth.

We also performed stratified analyses by cohort of arrival to Ontario and plotted the predicted

probabilities of preterm birth according to the duration of residence (Figure 3.3). Although

cohorts differed somewhat in their risk of preterm birth (more recent cohorts have higher risk;

this being consistent with the secular trends in preterm birth over time), the pattern of increasing

preterm birth with longer duration was pervasive across all four cohorts.

Although we based our covariate adjustment on causal diagrams, we also fitted our models with

the excluded variables material deprivation and infant sex but the results of Table 3.3 remained

unaffected. We expected that the inclusion of material deprivation would have attenuated the

association between duration and birth outcomes but it had the opposite effect, basically because

63

recent immigrants (who had the best outcomes) were more concentrated in deprived

neighbourhoods than long-term immigrants (who had the worse outcomes). As maternal

education, marital status, and knowledge of official languages were measured at arrival, and

therefore affected by the age at arrival, we repeated our analyses restricted to women aged 20

years and more at arrival but the association between duration of residence and adverse birth

outcomes did not change substantially. Finally, although the possibility of confounding by

secular increases in preterm delivery was controlled for by design (i.e., we restricted the study

population to liveborns delivered in the most recent 5-year period available), preterm deliveries

increased nearly 10% from fiscal year 2002/2003 to 2006/2007, thus raising the possibility of

residual confounding by trends over time within the study period. We therefore included year of

birth in the adjusted model to control for such potential residual confounding but its addition did

not have any visible impact on the effect estimates of duration of residence. Stratified analyses

restricted to one fiscal year at a time showed an independent association between duration of

residence and preterm birth, thus confirming that the association was not confounded by year of

delivery.

3.4. Discussion

The longer the time immigrant women resided in urban Ontario the higher their probability of

having adverse birth outcomes, except for small for gestational age. This exception, in

conjunction with a gestational age distribution shifted to the left among women with 15 or more

years spent in Canada and with the disappearance of the association between duration of

residence and LBW and its components after adjusting for preterm birth, strongly suggests that

the observed association between length of stay in Canada and adverse birth outcomes was

64

mainly driven by decreases in gestational age, which is consistent with a previous study on

Mexicans in the US.13 This finding cannot be explained by the secular trends in preterm birth

35,36 because we controlled such potential confounding effect by design (i.e., restriction of infants

born in the most recent 5-year period available, within which secular trends in the outcomes were

largely minimized). Confounding by cohort effects was a more serious threat that was partially

accounted for in our main analyses by adjusting for immigrant class and maternal country and

region of birth. We ruled out confounding by cohort effects as an alternate explanation to our

findings by repeating our analyses on a group of immigrant women arrived in the late ‘80s and

observed from 1988 to 2007 and the results were consistent with our main analyses (see

Appendix 3.E). Although cohort effects may have confounded somewhat our estimates,

principally those of particular world regions, the magnitude of such potential bias is small, since

the estimates obtained with the immigrants arrived from 1985 to 1989 were very similar to those

of the main analysis.

Our findings do not seem to support the convergence hypothesis. Low birthweight was

somewhat higher among recent immigrants compared with non-immigrants and continued to

increase with duration, contrary to the hypothesis. Recent immigrants were protected from

preterm birth and small preterm but lost their advantage after approximately 10 years. Instead of

remaining at that level, immigrants experienced a continuous deterioration that placed them at a

disadvantage after 15 years of stay, compared with non-immigrants. Rather than a ‘convergence’

the pattern of risk of preterm birth exhibited an ‘overshoot’, in which the risk of preterm birth

passed over the risk at which a plateau would be expected. Finally, because small for gestational

age was higher among immigrants than among non-immigrants a convergence would have

predicted a decrease but duration of residence had no visible impact on this outcome.

65

Our study had some limitations. Out of hospital births are not captured by the DAD. Gestational

age (in weeks) was measured in the DAD at the time of maternal admission but not at delivery

for the study period. Therefore, preterm birth rates may be higher using the DAD than other data

sources (e.g., Vital Statistics). It is unlikely that the bias resulting from using gestational age at

admission rather than at delivery would account for the observed association between duration of

residence and preterm birth, since maternal length of stay at the hospital from the admission date

to the delivery date showed little variation with duration of residence, and such variation was

explained by maternal sociodemographic characteristics (not shown).

Because immigrants’ data started in 1985 we could not identify those immigrants that obtained

their permanent residence prior to that year. This misclassification, however, would bias our

results towards the null because false non-immigrants were immigrants with at least 17 years of

residence, presumably at higher risk of adverse birth outcomes.

Although immigration data was of good quality, mostly ascertained by notarized documentation

provided by the applicants during the immigration process, some variables were measured at

arrival and not at delivery, which may be more relevant for time-dependent variables. Marital

status and maternal education may have changed for some women, especially for young women

at arrival who may have become married and more educated with longer duration of residence.

As educational attainment is not reversible the bias resulting from adjustment for an

underestimate of the true educational attainment of long term immigrants would have moved our

results towards the null, because higher education protects against adverse birth outcomes.

We could not control for maternal health behaviours during pregnancy, particularly tobacco

smoking. Although adjustment for these factors is not appropriate because they are

66

conceptualized as mediators between socioeconomic exposures and adverse birth outcomes,37

their consideration may have clarified potential mechanisms involved with duration of residence.

To measure small for gestational age we used a sex-specific Canadian standard (excluding

Ontario). 26 The use of a single standard of birthweight by gestational age that does not take into

account ethnic group differences may not be valid for infants born to mothers from particular

regions of the world, such as Asians. 38,39 The failure of using ethnic-specific standards of

birthweight by gestational age may have resulted in misclassification (i.e., greater probability of

classifying infants born to Asian mothers as SGA when they would not be SGA according to an

Asian-specific standard).39 Therefore, SGA findings must be interpreted with caution,

particularly when comparing immigrants versus non-immigrants. However, the effects estimates

of duration of residence on SGA within each migrant group defined by their maternal world

region of origin are supposed to be less biased because the reference group is internal to the

ethnic group (i.e., recent immigrants). It remains to be determined whether the use of ethnic-

specific standards may reveal an association between duration of residence and SGA among

various immigrant groups.

Despite these limitations, the association between duration of residence and gestational age-

related outcomes was quite robust, remaining unaltered under a range of sensitivity analyses,

such as covariate adjustment, restriction to women aged 20 years and older, and a different study

population composed of births delivered between 1988 to 2007.

Although our data were not detailed enough to elucidate mechanisms we can explore some clues.

First, the distribution of gestational age among immigrants with 15 or more years of stay is

shifted towards the left compared with the distribution of women delivering within the first five

years of stay. Although the curves do not reflect a true longitudinal process (i.e., the two groups

67

were composed of different women), it is noteworthy that the differences between the

distributions are spread all over the gestational age range. This feature strongly suggests the

presence of influences operating across the entire range of the gestational age distribution rather

than on a ‘high-risk’ subgroup affecting a critical period of gestation. 40,41

One candidate explanation is acculturation, which has already been linked with adverse birth

outcomes.6-8,42 Although a complex construct, acculturation basically entails a process by which

immigrants incorporate the behaviours and values of the host population.43 This suggests that the

effect of duration of residence may be mediated by changes in health behaviours and related risk

factors. Indeed, duration of residence has been used as a proxy for acculturation and been

associated with increases in obesity/overweight, smoking, alcohol consumption, and physical

inactivity, 17-20,44,45 factors that may negatively affect gestational age. Although we could not

assess mediation by these factors, secondary data suggest that acculturation is a plausible

hypothesis. Immigrants to Canada as a whole were less likely to have high-fat and carbohydrate

diet than non-immigrants46 as well as lower cigarette smoking and alcohol consumption

rates.47,48 Perhaps more intriguing is that immigrants with more than 10 years of residence in

Ontario had a right-shifted body weight distribution compared with those with less than 10 years

of stay, and twice their obesity rate. (Canadian Community Health Survey cycle 3.1, 2005)

While high pre-pregnancy weight has not been consistently associated with preterm birth in the

literature,49-51 it is interesting to note that higher body mass index has been found to be protective

for small for gestational age.49,52-54 It is reasonable to speculate that in our immigrant population

increases in body mass index may have compensated the detrimental effect of changes in risk

factors thus rendering a null association between duration of residence and small for gestational

age.

68

Another unexplored aspect of acculturation to Western societies may involve increased

medicalization with time spent in Canada. Such medicalization may be translated in a shortening

of the time from conception to first prenatal visit, increased compliance and enhanced prenatal

surveillance, preference for elective cesarean delivery, and use of assisted reproductive

technologies. When compared with menstrual dates, early ultrasound based dating results in a

left shift in the gestational age distribution. If immigrants with longer duration of residence were

more medicalized and enter antenatal care earlier, their gestational age estimate might has been

more likely to be based on early ultrasound than among less medicalized immigrant women who

had their first prenatal visit at a more advanced stage of their pregnancies. Even assuming

complete polarization of early ultrasound use in our study population, this could not completely

explain our results, since the bias resulting from using ultrasound dating only versus the last

menstrual period only was estimated to be no more than a 10-20% increase in the risk of preterm

birth,55 and we observed a 14%, 30%, and 49% increased odds associated with 5, 10, and 15

years of residence, respectively. Another potential pathway related to increased medicalization

may be increased preference for elective cesarean section, which may shorten gestation length

for some women. It is also possible that use of assisted reproductive technologies (ARTs) have

become more frequent among the more acculturated migrant couples. Use of ARTs has been

linked with increased risk of preterm birth, even among singleton infants.56

Another potential pathway may be maternal morbidity during pregnancy, especially pregnancy-

induced hypertension and incompetent cervix, conditions that increased with duration of

residence. However, these trends were counterbalanced by declining rates of genito-urinary

infections, which inclusion in the models strengthened the association between duration of

residence and preterm birth.

69

A last potential pathway leading to adverse birth outcomes, albeit unexplored among immigrant

women, may be related to their labour market experiences. Working conditions, such as long

working hours, prolonged standing, and physically demanding work has been associated with

PTB.57,58 Psychosocial exposures such as job strain, low job control and satisfaction may be

concomitant causes of preterm birth among immigrant women,59-62 who are more likely to be

employed in manual, clerical, and shift jobs. Despite 1990’s immigration policies favouring

immigrants with higher education, which resulted in an almost doubling of the rate of recent

immigrants with bachelor’s degrees to more than 40% in 2001, one third of recent immigrant and

one fourth of other immigrant working women worked in jobs requiring less than their education

level.63 Moreover, immigrants are more likely than non-immigrants to live in deprived

neighbourhoods, which may contribute to ‘weather’ migrant women, jointly with experiences of

subordination and discrimination.64

The clarification of the mechanisms behind the association between duration of residence and

preterm birth merits further investigation, especially when the contribution of immigrants to live

births in developed countries is growing. Our findings suggest that influences at the population

level are likely to play a key role and support the use of length of residence as an important

indicator for surveillance of immigrant health.

70

3.5. References



(2) Bona G, Zaffaroni M, Cataldo F et al. Infants of immigrant parents in Italy. A national multicentre case

control study. Panminerva Medica. 2001;43(3):155-159.





(4) Stoltenberg C, Magnus P, Stoltenberg C, Magnus P. Children with low birth weight and low gestational age

in Oslo, Norway: immigration is not the cause of increasing proportions. Journal of Epidemiology &

Community Health. 1995;49(6):588-593.



(6) Cobas JA, Balcazar H, Benin MB, Keith VM, Chong Y. Acculturation and low-birthweight infants among

Latino women: a reanalysis of HHANES data with structural equation models. Am J Public Health.

1996;86(3):394-396.


Public Health. 1989;79(9):1263-1267.

71






Int J Gynaecol Obstet. 2003;83(3):257-265.

(11) Singh GK, Yu SM. Adverse pregnancy outcomes: differences between US- and foreign-born women in

major US racial and ethnic groups. Am J Pub Health. 1996;86(6):837-843.

(12) Crump C, Lipsky S, Mueller BA, Crump C, Lipsky S, Mueller BA. Adverse birth outcomes among

Mexican-Americans: are US-born women at greater risk than Mexico-born women? Ethnicity & Health.

1999;4(1-2):29-34.



(14) Rasmussen F, Oldenburg CE, Ericson A et al. Preterm birth and low birthweight among children of

Swedish and immigrant women between 1978 and 1990.[erratum appears in Paediatr Perinat Epidemiol


(15) Abraido-Lanza AF, Armbrister AN, Florez KR, Aguirre AN. Toward a theory-driven model of

acculturation in public health research. Am J Public Health. 2006;96(8):1342-1346.

72

(16) Syme SL, Marmot MG, Kagan A, Kato H, Rhoads G. Epidemiologic studies of coronary heart disease and

stroke in Japanese men living in Japan, Hawaii and California: introduction. Am J Epidemiol.

1975;102(6):477-480.

(17) Goel MS, McCarthy EP, Phillips RS, Wee CC. Obesity among US immigrant subgroups by duration of

residence. JAMA. 2004;292(23):2860-2867.

(18) Hosper K, Nierkens V, Nicolaou M, Stronks K. Behavioural risk factors in two generations of non-Western

migrants: do trends converge towards the host population? Eur J Epidemiol. 2007;22(3):163-172.

(19) Choi S, Rankin S, Stewart A, Oka R. Effects of acculturation on smoking behavior in Asian Americans: a

meta-analysis. J Cardiovasc Nurs. 2008;23(1):67-73.

(20) Mejean C, Traissac P, Eymard-Duvernay S, Delpeuch F, Maire B. Influence of acculturation among

Tunisian migrants in France and their past/present exposure to the home country on diet and physical

activity. Public Health Nutr. 2008;1-10.

(21) Bos V, Kunst AE, Garssen J, Mackenbach JP. Duration of residence was not consistently related to

immigrant mortality. J Clin Epidemiol. 2007;60(6):585-592.

(22) Statistics Canada. Immigration in Canada: A Portrait of the Foreign-born. Census year 2006. Catalogue no.

97-557-XIE. 2007. Ottawa, Statistics Canada, Minister of Industry.



(24) Statistics Canada. Standard Geographical Classification (SGC). Volume I. The Classification. [Statistics

Canada]. 2007. Accessed December 12, 2008.

73

(25) Wilcox AJ. On the importance--and the unimportance--of birthweight. Int J Epidemiol. 2001;30(6):1233-

1241.

(26) Kramer MS, Platt RW, Wen SW et al. A new and improved population-based Canadian reference for birth

weight for gestational age. Pediatrics. 2001;108(2):E35.

(27) UNICEF. Countries by Region. [http://www unicef org/infobycountry/index html]. 2008.

(28) Matheson FI, Moineddin R, Dunn JR, Creatore MI, Gozdyra P, Glazier RH. Urban neighborhoods, chronic

stress, gender and depression. Soc Sci Med. 2006;63(10):2604-2616.

(29) Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology.

1999;10(1):37-48.

(30) Jewel NP. Statistics in Epidemiology. Boca Raton: Chapman & Hall/CRC; 2004.

(31) Diez-Roux AV. Bringing context back into epidemiology: variables and fallacies in multilevel analysis. Am

J Public Health. 1998;88(2):216-222.


2002.

(33) Branum AM, Schoendorf KC. Changing patterns of low birthweight and preterm birth in the United States,

1981-98. Paediatr Perinat Epidemiol. 2002;16(1):8-15.

(34) Public Health Agency of Canada. Canadian Perinatal Health Report, 2008 Edition. Public Health Agency

of Canada . 2008. Ottawa, Public Health Agency of Canada.

74

(35) Ananth CV, Joseph KS, Oyelese Y, Demissie K, Vintzileos AM. Trends in preterm birth and perinatal

mortality among singletons: United States, 1989 through 2000. Obstet Gynecol. 2005;105(5 Pt 1):1084-

1091.

(36) Kramer MS, Platt R, Yang H et al. Secular trends in preterm birth: a hospital-based cohort study. JAMA.

1998;280(21):1849-1854.

(37) Kramer MS, Goulet L, Lydon J et al. Socio-economic disparities in preterm birth: causal pathways and

mechanisms. Paediatr Perinat Epidemiol. 2001;15 Suppl 2:104-123.

(38) Kierans WJ, Joseph KS, Luo ZC, Platt R, Wilkins R, Kramer MS. Does one size fit all? The case for

ethnic-specific standards of fetal growth. BMC Pregnancy Childbirth. 2008;8:1.

(39) Ray J, Jiang D, Sgro M, Shah R, Singh G, Mamdani M. Thresholds for Small for Gestational Age Among

Newborns of East Asian and South Asian Ancestry. J Obstet Gynaecol Can. 2009;31(4):322-330.

(40) Rose G. Sick individuals and sick populations. Int J Epidemiol. 2001;30(3):427-432.

(41) Tunstall-Pedoe H, Connaghan J, Woodward M, Tolonen H, Kuulasmaa K. Pattern of declining blood

pressure across replicate population surveys of the WHO MONICA project, mid-1980s to mid-1990s, and

the role of medication. BMJ. 2006;332(7542):629-635.

(42) Hyman I, Dussault G. The effect of acculturation on low birthweight in immigrant women. Can J Public

Health. 1996;87(3):158-162.

(43) Lara M, Gamboa C, Kahramanian MI, Morales LS, Bautista DE. Acculturation and Latino health in the

United States: a review of the literature and its sociopolitical context. Annu Rev Public Health.

2005;26:367-397.

75

(44) Kaplan MS, Huguet N, Newsom JT, McFarland BH. The association between length of residence and

obesity among Hispanic immigrants. Am J Prev Med. 2004;27(4):323-326.

(45) Park Y, Neckerman KM, Quinn J, Weiss C, Rundle A. Place of birth, duration of residence, neighborhood

immigrant composition and body mass index in New York City. Int J Behav Nutr Phys Act. 2008;5:19.

(46) Pomerleau J, Ostbye T, Bright-See E. Place of birth and dietary intake in Ontario. I. Energy, fat,

cholesterol, carbohydrate, fiber, and alcohol. Prev Med. 1998;27(1):32-40.

(47) Millar WJ, Hill G. Pregnancy and smoking. Health Rep. 2004;15(4):53-56.

(48) McDonald JT, Kennedy S. Insights into the 'healthy immigrant effect': health status and health service use

of immigrants to Canada. Soc Sci Med. 2004;59(8):1613-1627.

(49) Cnattingius S, Bergstrom R, Lipworth L, Kramer MS. Prepregnancy weight and the risk of adverse

pregnancy outcomes. N Engl J Med. 1998;338(3):147-152.

(50) Committee on Understanding Premature Birth and Assuring Healthy Outcomes. Preterm Birth: Causes,

Consequences, and Prevention. First ed. Washington, D.C.: National Academies Press; 2007.

(51) Nohr EA, Bech BH, Vaeth M, Rasmussen KM, Henriksen TB, Olsen J. Obesity, gestational weight gain

and preterm birth: a study within the Danish National Birth Cohort. Paediatr Perinat Epidemiol.

2007;21(1):5-14.

(52) Baeten JM, Bukusi EA, Lambe M. Pregnancy complications and outcomes among overweight and obese

nulliparous women. Am J Public Health. 2001;91(3):436-440.

76

(53) Cheng CJ, Bommarito K, Noguchi A, Holcomb W, Leet T. Body mass index change between pregnancies

and small for gestational age births. Obstet Gynecol. 2004;104(2):286-292.

(54) Lu GC, Rouse DJ, DuBard M, Cliver S, Kimberlin D, Hauth JC. The effect of the increasing prevalence of

maternal obesity on perinatal morbidity. Am J Obstet Gynecol. 2001;185(4):845-849.

(55) Henriksen TB, Wilcox AJ, Hedegaard M, Secher NJ. Bias in studies of preterm and postterm delivery due

to ultrasound assessment of gestational age. Epidemiology. 1995;6(5):533-537.

(56) Schieve LA, Ferre C, Peterson HB, Macaluso M, Reynolds MA, Wright VC. Perinatal outcome among

singleton infants conceived through assisted reproductive technology in the United States. Obstet Gynecol.

2004;103(6):1144-1153.

(57) Bonzini M, Coggon D, Palmer KT. Risk of prematurity, low birthweight and pre-eclampsia in relation to

working hours and physical activities: a systematic review. Occup Environ Med. 2007;64(4):228-243.

(58) Mozurkewich EL, Luke B, Avni M, Wolf FM. Working conditions and adverse pregnancy outcome: a

meta-analysis. Obstet Gynecol. 2000;95(4):623-635.

(59) Oths KS, Dunn LL, Palmer NS. A prospective study of psychosocial job strain and birth outcomes.

Epidemiology. 2001;12(6):744-746.

(60) Gisselmann MD, Hemstrom O. The contribution of maternal working conditions to socio-economic

inequalities in birth outcome. Soc Sci Med. 2008;66(6):1297-1309.

(61) Croteau A, Marcoux S, Brisson C. Work activity in pregnancy, preventive measures, and the risk of

preterm delivery. Am J Epidemiol. 2007;166(8):951-965.

77

(62) Saurel-Cubizolles MJ, Zeitlin J, Lelong N, Papiernik E, Di Renzo GC, Breart G. Employment, working

conditions, and preterm birth: results from the Europop case-control survey. J Epidemiol Community

Health. 2004;58(5):395-401.

(63) Galarneau D, Morissette R. Immigrants: Settling for Less? Perspectives on Labor and Income. 2004;5(6):5-

16.

(64) Rich-Edwards JW, Grizzard TA. Psychosocial stress and neuroendocrine mechanisms in preterm delivery.

Am J Obstet Gynecol. 2005;192(5 Suppl):S30-S35.

78

Chapter 4 The Interplay between Immigrants’ Country of Birth and

Neighbourhood Deprivation on Birth Outcomes

Abstract

Objectives: We compared the influence of the residential environment and of maternal country

of birth on the birthweight and low birthweight of infants born to recent immigrants to urban

Ontario.

Methods: We linked delivery records (1993-2000) to an immigration database (1993-1995) and

small-area census data (1996). The data were analyzed using cross-classified random effects

models (CCREM) and standard multilevel methods. Higher-level predictors included four

independent measures of neighbourhood context constructed using factor analysis, and maternal

world regions of origin.

Results: Births (N=22,189) were distributed across 1,396 census tracts and 155 countries of

origin. The associations between neighbourhood indices and birthweight disappeared after

controlling for the maternal country of birth in a cross-classified multilevel model. Significant

associations between world regions and birthweight and low birthweight persisted after

controlling for the neighbourhood context and individual characteristics.

Conclusions: Residential environment has little, if any, influence on birthweight among recent

immigrants to Ontario. Regarding low birthweight, the origin of recent immigrants is more

important than where they currently live. Caution should be exercised when interpreting findings

of neighbourhood influences on perinatal outcomes among recent immigrants.

79

4.1. Introduction

Socioeconomic disparities in birth outcomes are well documented,1-3 even in countries with

universal access to health care 4-6 such as Canada. There is also an increasing body of literature

suggesting that context affects birth outcomes, particularly neighbourhood influences in

predominantly urban areas, including several multilevel studies.7-18

Little is known, however, about neighbourhood influences among immigrants.12,19-21 There are

theoretical and practical reasons to explore this issue. It has been suggested that exposure to

neighbourhoods may take some time to exert its effects on human health.22 Even if

neighbourhood influences are detected among the offspring of recent-immigrant women exposed

to neighbourhoods during their entire pregnancy, a life-course perspective suggests that early life

experiences and pre-migration exposures may still affect birth outcomes of migrants in the new

country.19,22,23 The maternal country of birth thus constitutes another relevant “context” to be

considered when analyzing differences in birthweight among recent immigrants, since

substantial differences in birthweight have been reported by geographical region and nativity

status.24-26 It is important to clarify the role of the pre- and post-migration exposures, since the

proportion of live births to immigrant women has been showing an upward trend during recent

decades in several industrialized countries.24,27

The purpose of our study is to compare the influence of the residential environment at the time of

delivery, with that of the maternal country of birth, on the birthweight and low birthweight of

infants born to recent immigrant women who settled in Ontario census metropolitan areas from

1993 to 1995. We hypothesized that the maternal country of birth will have a greater impact on

their infant’s birthweight than the residential environment in which immigrants currently reside

in urban Ontario.

80

4.2. Methods

4.2.1. Data

Anonymized birth and maternal obstetric records from all Ontario hospitals were extracted from

hospital discharge abstracts compiled by the Canadian Institute for Health Information (1993-

2000). These data were internally merged to combine maternal and newborn records using an

algorithm described elsewhere,28 resulting in a 95 percent valid match of the newborn records to

a mother. Encrypted health care numbers of the Ontario Health Insurance Plan, which provides

universal access to nearly all physician and hospital services (except for asylum-seekers and

before three months’ residence), were used to link birth data with the Landed Immigrant Data

System (LIDS) (1993-1995), compiled by Citizenship and Immigration Canada. The LIDS

contains sociodemographic information and characteristics related to the immigration process.

These data were finally merged with small area data from the 1996 Canadian census.

We selected a group of women immigrating to Ontario’s census metropolitan areas from January

1, 1993 to March 31, 1995 who had at least one live singleton weighing more than 500 grams

and less than 6,000 grams (N= 38,121). We restricted the study population to infants born to

recent immigrant mothers, defined as residents with less than five years of stay.29,30 We slightly

modified this definition by shifting the five year observation period to begin after the 40th week

of the mothers’ arrival, to ensure that all mothers had been exposed to Canadian neighbourhoods

during their entire pregnancy (N=29,625). We only retained the first Canadian singleton born

alive of each woman within the five-year period (N= 22,516). The study population used for

analyses consisted of 22,189 live singleton infants born to women who immigrated to the

Ontario census metropolitan areas from January 1993 to March 1995, after excluding births to

mothers aged less than 15 years and more than 55 years (N=5), with atypical immigrant class

81

(N=62), coming from countries that could not be classified according to their socioeconomic

conditions (N=180) and records with missing information on place of residence (N=80).

Mothers born in one of 155 countries were distributed across 1,396 census tracts from ten 1996

Ontario Census Metropolitan Areas.31 Census tracts (our neighbourhoods) are relatively stable

urban neighbourhoods with a typical population of 2,500 to 8,000 and are relatively

homogeneous with respect to population characteristics and living conditions.31 Use of these data

was approved within the ethical review process of the Sunnybrook Health Sciences Centre,

Toronto, Ontario.

4.2.2. Outcomes

Birthweight was modeled as a continuous response variable (measured in grams). Use of

birthweight as a continuous outcome has several advantages. Birthweight is measured with high

precision, and provides directly interpretable effect estimates (differences expressed in grams)

and greater statistical power relative to categorical outcomes. We also modeled low birthweight,

as a binary outcome (proportion of births weighting less than 2,500 grams). It has been argued

that birthweight is a more meaningful perinatal outcome (i.e., proxy for fetal growth) if restricted

to term births, since the mean is not affected by the residual distribution (i.e., left-tale of the

birthweight distribution, mainly composed of preterm infants). 32 We did include all births

because the focus of this paper is not on fetal growth and the use of the same outcome definition

ensures comparability of our findings with those already published in the literature of

neighbourhood and contextual effects on perinatal health. 3,7,8,13,14,17,21 Birthweight (and low

birthweight) are sensitive to environmental influences and these influences are the focus of this

paper. Sensitivity analyses using birthweight at term and preterm birth did not alter the main

findings (not shown).

82

4.2.3. Predictors

Table 4.1 presents the predictors at each level of the hierarchical structure of the data. To obtain

groups of countries as homogeneous as possible in terms of their socioeconomic conditions, we

considered world region and the income-level of the country of birth. Both variables are based on

the World Bank classification of world economies 2000,33 in which countries are classified

according to their gross national income (GNI) per capita, using the World Bank Atlas method.

We modified the World Bank sub-region classification by separating the United States from the

remaining countries of the Americas. The group with the highest mean birthweight (East

Europe/Central Asia) was used as the referent.

We used more than one indicator of neighbourhood context based on the literature linking

neighbourhood stressors and poor health.14,34-37 In order to capture the complexity of the

neighbourhood environment, four independent measures were obtained by factor-analyzing

census variables; this had the added benefit of avoiding problems of multicollinearity, since the

principal components are not correlated. The neighbourhood indices are material deprivation

(Cronbach’s alpha 0.88), residential instability (Cronbach’s alpha 0.93), dependency

(Cronbach’s alpha 0.72) and ethnic diversity (Cronbach’s alpha 0.93). All four indices are

standardized continuous scores for modeling purposes. They are collapsed into tertiles for

descriptive purposes in Table 4.1. Details of how these measures were constructed and their

statistical properties are given elsewhere.38

At the individual level we considered established predictors of birthweight available in our data,

and some circumstances of the immigration process that may operate as potential confounders.

We considered: infant sex (male as referent), maternal age (15-19, 20-24, 25-29, 30-34 as

referent, and 35-55 years), maternal education groups (0-9, 10-12 years, some post-secondary

83

non-university diploma, and university diploma as referent), and marital status (married as

referent, single/widowed/separated). Gestational age in weeks was not available in the discharge

records for the study period. We approximated gestational age using gestational age groups (less

than 28, 28-36, 37-41 as referent and 42 and more weeks), based on the International

Classification of Diseases codes (ninth revision).39 Circumstances of immigration that have been

linked to birth outcomes and may vary by maternal country of birth include immigrant class

(economic class, family class, with refugees as referent) 40,41 and self-reported knowledge of any

official Canadian language (English or French) (”yes” as referent). 42,43 Length of residence in

Canada after the beginning of the observation period (0 to 4 completed years) was used to assess

the amount of exposure to a Canadian setting. 29,44,45


Commonly used multilevel models cannot be used when the data do not present a purely nested

structure. Immigrant mothers living in a particular neighbourhood may have come from several

different countries, and mothers coming from a particular country may settle in different

neighbourhoods. Thus, this data structure presents a cross-classification of countries and

neighbourhoods. Raudenbush and Bryk developed an extension of the multilevel model to

analyze such data, known as the crossed-classified random effects model (CCREM). 46,47 (see

Appendix 4.A for technical details) According to guidelines based on simulation studies, the

number of units at each level of our data structure is sufficient to obtain unbiased and precise

regression coefficients, variance components and standard errors. 48,49

84

4.2.5. Modeling strategy

Prior to developing the cross-classified model we conducted preliminary analyses focusing on

one level at a time 50 to assess whether there was significant variation at each level separately.

We first used the usual two-level random intercept model with births as level-1 units and with

neighbourhood as level-2 units. In a second model countries were the level-2 units. The presence

of statistically significant variance for each of these models warrants the use of the cross-

classified model to assess whether the variations in the outcome at each level are independent (in

which case further modeling is based on the CCREM) or are associated (if one factor is

confounded by the other, rendering a variance non-significant, further modeling may be reduced

to the usual two-level model). Then we proceeded to sequentially fit models adjusted for

individual-level characteristics and group-level characteristics. We also tested for cross-level

interactions to see whether the effect of neighbourhood-level variables differed by maternal

country of birth but none were statistically significant.

The MIXED procedure in SAS 9.1 for UNIX was used to fit models using the continuous

measure of birthweight and PROC GLIMMIX was used to model low birthweight (SAS Institute

Inc., Cary, NC). Variance components estimates are reported with their standard errors and p-

values. The proportion of variance explained at each level was calculated with the intra-class

correlation coefficient (ICC). 46-49 Wald tests were used to test for significant variances in low

birthweight. 48 Fixed effects in birthweight are reported as differences in mean birthweight

expressed in grams; fixed effects in low birthweight are reported as adjusted odds ratios (OR)

with 95 percent confidence intervals (95% CI).

85

Table 4.1. Characteristics of the study population by geography, mean infant’s birthweight and low

birthweight among recent immigrant mothers to urban Ontario.

Countries

Census

tracts Births BW

Low BW

N (%) N (%) N (%) Mean 95% CI % 95% CI

Total 155 (100) 1396 (100) 22189 (100) 3288 (3281, 3295) 5.7 (5.4, 6.0)

Country-level characteristics World regions East Europe / Central Asia 25 (16.1) 730 (52.3) 2188 (9.9) 3497 (3475, 3520) 2.9 (2.2, 3.6) Rest of Europe 22 (14.2) 589 (42.2) 1022 (4.6) 3421 (3390, 3453) 3.8 (2.7, 5.0) United States 1 ( 0.6) 319 (22.9) 411 (1.8) 3463 (3413, 3513) 2.4 (0.9, 3.9) Latin America & Caribbean 32 (20.6) 829 (59.4) 3467 (15.6) 3265 (3245, 3285) 7.4 (6.5, 8.3) East Asia / Pacific 18 (11.6) 995 (71.3) 5880 (26.5) 3237 (3225, 3250) 5.4 (4.8, 5.9) South Asia 6 ( 3.9) 793 (56.8) 5960 (26.9) 3207 (3194, 3220) 7.0 (6.3, 7.6) Middle East 12 ( 7.7) 516 (37.0) 1285 (5.8) 3359 (3332, 3385) 3.7 (2.7, 4.8) North Africa 7 ( 4.5) 160 (11.5) 202 (0.9) 3435 (3364, 3505) 3.5 (0.9, 6.0) East/South Africa 19 (12.3) 466 (33.4) 1328 (6.0) 3367 (3336, 3397) 4.8 (3.8, 6.0) West Africa 13 ( 8.4) 202 (14.5) 446 (2.0) 3220 (3158, 3282) 9.2 (6.5, 11.9) <0.001a <0.001b Country income level Low Income 47 (30.3) 978 (70.1) 6691 (30.2) 3234 (3221, 3246) 6.4 (5.8, 7.0) Lower Middle Income 43 (27.7) 1146 (82.1) 10308 (46.5) 3293 (3283, 3304) 5.9 (5.4, 6.4) Upper Middle Income 31 (20.0) 874 (62.6) 2614 (11.8) 3348 (3327, 3369) 4.6 (3.8, 5.4) High Income non-OECD 10 ( 6.5) 361 (25.9) 1153 (5.2) 3256 (3229, 3283) 4.5 (3.3, 5.7) High Income (OECD) 24 (15.5) 778 (55.7) 1423 (6.4) 3418 (3392, 3445) 3.4 (2.4, 4.3) <0.001c <0.001d Neighbourhood-level characteristics

Material deprivation tertiles 1 Lowest 140 (90.3) 740 (53.0) 7336 (33.1) 3296 (3284, 3308) 5.1 (4.6, 5.6) 2 142 (91.6) 401 (28.7) 7413 (33.4) 3298 (3286, 3311) 5.6 (5.1, 6.1) 3 Highest 126 (81.3) 255 (18.3) 7440 (33.5) 3270 (3257, 3282) 6.4 (5.8, 6.9) 0.003c <0.001d Residential instability tertiles 1 Lowest 136 (87.7) 610 (43.7) 7387 (33.3) 3268 (3257, 3280) 5.6 (5.1, 6.1) 2 139 (89.7) 430 (30.8) 7358 (33.2) 3294 (3281, 3306) 6.0 (5.5, 6.6) 3 Highest 142 (91.6) 356 (25.5) 7444 (33.5) 3302 (3289, 3314) 5.4 (4.9, 5.9) <0.001c 0.586d Dependency tertiles 1 Lowest 144 (92.9) 464 (33.3) 7382 (33.3) 3305 (3293, 3318) 5.5 (5.0, 6.1) 2 137 (88.4) 401 (28.7) 7373 (33.2) 3271 (3259, 3284) 5.9 (5.3, 6.4) 3 Highest 135 (87.1) 531 (38.0) 7434 (33.5) 3287 (3275, 3299) 5.6 (5.1, 6.1) 0.038c 0.914d Ethnic diversity tertiles 1 Lowest 143 (92.3) 864 (61.9) 7367 (33.2) 3316 (3304, 3328) 5.1 (4.6, 5.6) 2 137 (88.4) 336 (24.1) 7405 (33.4) 3280 (3267, 3292) 6.0 (5.4, 6.5) 3 Highest 131 (84.5) 196 (14.0) 7417 (33.4) 3268 (3256, 3280) 5.9 (5.4, 6.5) <0.001c 0.027d Individual-level characteristics

Infant sex Male 11357 (51.1) 3337 (3327, 3347) 5.4 (4.9, 5.8) Female

10832 (49.9) 3236 (3227, 3246) 6.0 (5.5, 6.4)

Gestational age (completed weeks)

< 28 98 (0.4) 842 (779, 906) 99.0 (99.0, 100.0) 28-36 1016 (4.6) 2353 (2319, 2386) 62.5 (59.5, 65.5) 37-41 20949 (94.4) 3344 (3337, 3350) 2.5 (2.3, 2.7) ≥ 42 126 ( 0.6) 3494 (3407, 3581) 0.8 (0.0, 4.3) <0.001c <0.001d Maternal age group (years) 15-19 461 (2.1) 3188 (3139, 3237) 7.4 (5.0, 9.8) 20-24 3770 (17.0) 3221 (3205, 3237) 6.4 (5.6, 7.2) 25-29 7266 (32.7) 3289 (3277, 3301) 5.1 (4.6, 5.6) 30-34 6903 (31.1) 3323 (3310, 3336) 5.3 (4.8, 5.8) 35-54 3789 (17.1) 3301 (3283, 3320) 6.4 (5.7, 7.2) <0.001a 0.002b Maternal education 0 to 9 years of schooling 4100 (18.5) 3259 (3243, 3275) 6.3 (5.6, 7.1) 10 to 12 years of schooling 8339 (37.6) 3264 (3253, 3276) 6.2 (5.7, 6.7) Post-secondary non-university diploma

6152 (27.7) 3319 (3306, 3332) 5.0 (4.5, 5.6)

University diploma

3598 (16.2) 3323 (3305, 3340) 4.8 (4.1, 5.5)

86

Countries

Census

tracts Births BW

Low BW

N (%) N (%) N (%) Mean 95% CI % 95% CI

Marital status Married/common law 13764 (62.0) 3312 (3303, 3321) 5.3 (4.9, 5.7) Single/divorced/separated 8425 (38.0) 3248 (3237, 3260) 6.3 (5.8, 6.8) <0.001a 0.001b Immigrant class Economic class 4965 (22.4) 3325 (3310, 3341) 5.6 (5.0, 6.2) Family class 14137 (63.7) 3257 (3248, 3265) 5.9 (5.5, 6.3) Refugees 3087 (13.9) 3371 (3352, 3391) 4.9 (4.1, 5.6) <0.001a 0.09b Knowledge of English or French

Yes 13186 (59.4) 3302 (3293, 3312) 5.8 (5.4, 6.2) No 9003 (40.4) 3267 (3257, 3278) 5.5 (5.0, 5.9) <0.001a 0.29b Length of residence (completed years)

0 7734 (34.9) 3278 (3266, 3290) 5.4 (4.9, 5.9) 1 5028 (22.7) 3282 (3267, 3297) 6.0 (5.3, 6.6) 2 3949 (17.8) 3306 (3289, 3322) 5.1 (4.5, 5.8) 3 3016 (13.6) 3285 (3266, 3306) 6.2 (5.4, 7.1) 4 2462 (11.1) 3304 (3282, 3326) 6.1 (5.2, 7.0) 0.019c 0.148d

a p-value of analysis of variance used for comparison of means b p-value of Chi-Square test for comparison of proportions c p-value of linear-trend across means d p-value of the Cochran-Armitage test for trend for proportions

4.3. Results

The largest share of births was among women from Asian countries, followed by immigrants

from Latin American and Caribbean countries (Table 4.1). There were significant differences in

birthweight according to world regions and to the income level of the mother’s country of birth.

Differences in birthweight by neighbourhood tertiles were not very pronounced. Material

deprivation was the only neighbourhood characteristic showing a gradient in low birthweight in

the expected direction, such that higher material deprivation was associated with lower

birthweight.

At the individual-level, heavier birthweight was found for males and in infants with higher

gestational age (Table 4.1). Maternal characteristics associated with higher birthweight include:

increasing age up to 30-34 years although this differs for low birthweight, having a university

87

diploma, being married, being a refugee, having knowledge of English or French (although this

is reversed for low birthweight), and having spent more time in Canada although this differs for

low birthweight.

Table 4.2. Fixed effects (and 95% CI) of the neighbourhood indices on infant’s birthweight (in grams) and

random effects (and standard errors) among recent immigrants to urban Ontario.

Two level models with random neighbourhood variance Cross-classified models with random neighbourhood and

country of birth variances

Model 1: with

Neighbourhood predictors

Model 2: adjusted for

individual characteristics c

Model 3: with

Neighbourhood predictors

Model 4: adjusted for

individual characteristics c

Fixed effects beta (95% CI) beta (95% CI) beta (95% CI) beta (95% CI)

Intercept 3321 (3311, 3332) 3531 (3505, 3557) 3364 (3337, 3390) 3523 (3489, 3558)

Material deprivation -10 (-21, 1) 1 (-8, 10) -8 (-18, 2) -2 (-11, 7)

Residential instability 20 (10, 29) 12 (3, 20) 0 (-9, 9) -1 (-9, 7)

Dependency -14 (-24, -3) -9 (-18, 0) -8 (-18, 2) -7 (-15, 2)

Ethnic diversity -25 (-34, -17) -23 (-30, -16) -4 (-12, 4) -4 (-10, 3)

Random effects Variance

Standard

Error Variance

Standard

Error Variance

Standard

Error Variance

Standard

Error

: Variance at the

neighbourhood level 2022 b 571 1320 a 431 215 399 255 304

: Variance at the

country level 13737 b 2596 10400 b 1995

: Residual variance 283014 b 2724 208223 b 2007 272954 b 2626 201375 b 1939

a p< 0.01; b p<0.001 c Adjusted for infant sex, maternal age groups, gestational age groups, maternal education groups, immigrant class, marital status,

knowledge of English or French, length of residence in Canada.

Table 4.2 shows the results of the multilevel models assessing neighbourhood effects on

birthweight, before (Models 1 and 2) and after including the country-of-origin context (Models 3

and 4). Model 1 and 2 represent the usual two-level model with births nested within

neighbourhoods. Model 1 included the four neighbourhood factors, which were significant, with

88

the exception of the material deprivation score. All four neighbourhood factors explained 42

percent of the variability found across neighbourhoods. When individual characteristics were

included in Model 2 significant variability between neighbourhoods remained.

Models 3 and 4 show the results of the cross-classified models. They differ from Models 1 and 2

by including an additional random variance component at the country-level. The addition of the

country-level context in Model 3 rendered both the neighbourhood-level variance and all the

neighbourhood indices non-significant. The country-level variance, however, remained highly

significant even after adjusting for individual characteristics in Model 4. In this model, the

partition of the variance indicates that 4.9 percent of the total variance in birthweight occurred at

the country level and only 0.12 percent at the neighbourhood level. Collapsing the

neighbourhood indices into tertiles did not improve the fit of the model (data not shown) and

therefore we kept the continuous specification of the neighbourhood indices. Models adjusted for

individual characteristics show higher infant mean birthweight because most reference categories

are those associated with higher birthweight, such as being male infants born within 37-41

completed weeks of gestation, and born to married mothers with high maternal age and

education.

In order to understand why the variability in birthweight at the neighbourhood-level disappeared

after controlling for country of birth, we tested the hypothesis that the pattern of settlement of

recent immigrants was not random, by cross tabulating material deprivation tertiles of births with

world regions (Figure 4.1). The Chi-Square test was highly significant (χ2(df=18)=1,052,

p<0.0001), indicating that newly arrived immigrant women from particular regions of the world

did not settle randomly across urban neighbourhoods. Women coming from poorer regions of the

world settled in neighbourhoods characterized by higher material deprivation (e.g., African and

89

Latin American countries, and to a lesser extent South Asia) whereas women coming from richer

regions tended to concentrate in less deprived neighbourhoods (e.g., United States, non-Eastern

Europe, and East Asia and Pacific to a lesser extent). The neighbourhood-level variance and

predictors were no longer significant in the cross-classified models; the model then reduced to a

usual two-level model, with births nested within countries of origin.

Figure 4.1. Distribution of births by neighbourhood material deprivation tertiles in each world sub-region

Table 4.3 shows the results of the two-level model including world regions as the only country-

level predictor (Model 5), since income levels of the country of birth were no longer significant

after including world regions. The ICC indicates that 46.7 percent of the variance at the country

level was explained by grouping the countries into world regions. Significant differences in

90

birthweight between world regions persisted after controlling for individual characteristics in

Model 6, as did the unexplained variability at the country-level. Most world regions had lower

mean birthweight than East Europe/Central Asia, except the United States and North Africa,

probably due to low statistical power as a result of very few countries in these world regions.

Table 4.3. Fixed effects (and 95% CI) of world regions on infant’s birthweight (in grams) among recent

immigrants to urban Ontario.

Model 5: Unadjusted for

individual characteristics

Model 6: Adjusted for

individual characteristics a

World regions beta (95% CI) beta (95% CI)

East Europe / Central Asia referent referent

Rest of Europe -73 (-153, 7) -76 (-143, -9)

U.S.A. -42 (-229, 145) -47 (-200, 105)

Latin America & Caribbean -150 (-220, -80) -133 (-192, -75)

East Asia / Pacific -246 (-323, -169) -238 (-302, -174 )

South Asia -281(-377, -184 ) -241 (-320, -162)

Middle East -168 (-257, -79) -157 (-230, -83)

North Africa -83 (-207, 41) -89 (-192, 15)

East/South Africa -236 (-322, -151) -206 (-278, -135)

West Africa -226 (-350, -102 ) -195 (-298, -91)

a Adjusted for infant sex, maternal age groups, maternal education groups, gestational age groups, immigrant class,

marital status, knowledge of English or French, length of residence in Canada.

The results for low birthweight differed in the significance of the variance at the neighbourhood

level (Table 4.3). Unlike birthweight, there was no significant variability in low birthweight in

the two-level model with births nested within neighbourhoods ( = 0.01, standard error=0.03,

p > 0.05 one-sided) or in the cross-classified model considering both the neighbourhood and

country-level contexts. None of the neighbourhood indices were significantly associated with

91

low birthweight in either of these two models. We therefore dropped the neighbourhood context

and continued the modeling of low birthweight considering the country context only. Variability

in low birthweight at the country-level, in contrast, was significant in the two-level model with

births nested within countries and also in the cross-classified model ( = 0.20, standard

error=0.06, p < 0.001 one-sided). It remained significant after controlling for individual-level

covariates ( = 0.15, standard error=0.05, p < 0.01 one-sided), although the variance was

somewhat reduced.

Table 4.4. Odds ratios (and 95% CI) of world regions on infant’s Low Birthweight among recent

immigrants to urban Ontario.

World regions Odds ratio (95% CI) a

East Europe / Central Asia 1.00

Rest of Europe 1.59 (0.88, 2.86)

U.S.A. 0.87 (0.31, 2.43)

Latin America & Caribbean 2.15 (1.34, 3.45)

East Asia / Pacific 1.96 (1.21, 3.18)

South Asia 2.80 (1.67, 4.68)

Middle East 1.81 (1.02, 3.22)

North Africa 1.38 (0.49, 3.84)

East/South Africa 2.08 (1.17, 3.67)

West Africa 2.22 (1.01, 4.85) a Adjusted for infant sex, maternal age groups, maternal education groups, gestational age groups, immigrant class, marital status,

knowledge of English or French, length of residence in Canada.

The full model (Table 4.4) including all individual characteristics and world regions rendered the

country-level variance non significant ( = 0.07, standard error = 0.05, p > 0.05 one-sided),

implying that after grouping countries into regions there was no further variability to be

explained at the country-level. Fixed effect estimates resemble the pattern found for birthweight.

92

The risk of low birthweight varied considerably according to the region of origin of the

immigrant mothers.

4.4. Discussion

In a study population of recent immigrant women to urban Ontario, we found that the

neighbourhood context had little, if any, impact on birthweight and low birthweight.

Neighbourhood influences on birthweight disappeared after controlling for the mother’s country

of birth, suggesting that self-selection of recent immigrants from various world regions into

particular neighbourhoods explains the observed associations between neighbourhood

characteristics and birthweight. By contrast, we found important contextual effects at the

maternal country of birth level for both birthweight and low birthweight, after adjusting for

individual characteristics. Compared to migrants from East Europe/ Central Asia, migrants from

other world regions had worse outcomes, with the exception of North Africa, the United States

and Western Europe. Low rates of preterm birth (< 37 weeks of gestation) and low birthweight

have already been documented among North African migrants to Belgium 51-54 and France.55 US-

migrants to Ontario presented lower rates of singleton low birthweight than the US-born non-

Hispanic Whites in the US in a comparable period, 56 suggesting that US-migrants to Ontario

constitute a healthier group than their non-migrant counterparts.

Among the strengths of this study we highlight that this is a population-based study with almost

complete coverage of the target population. Selection bias is not an issue since almost all Ontario

permanent residents are insured by the provincial health plan. Unlike in many studies,

immigration status, country of birth, and other maternal characteristics were not self-reported but

93

ascertained through the governmental computerized immigration database, which is quite

accurate and complete because of its associated legal requirement for notarized copies of the

personal documentation of the principal applicant and family members. The use of an

appropriate statistical method that allows simultaneous consideration of the role of two relevant

contexts strengthens our conclusions.

This study has some limitations. Imperfect measurement of some individual control variables

may have introduced some residual confounding. Maternal education and marital status were

measured at arrival but could have changed for some women during the study period; however

the relatively short period of time from arrival to delivery makes it improbable that substantial

shifts in educational attainment were experienced by many women before delivering their first

Canadian-born child. Information on parity was not available in hospital records for the study

period. We reduced confounding for parity, however, by selecting only the first Canadian-infant

born of each mother, although many women may already have had prior babies. We do not have

reason to believe that parity would differ systematically according to the country of birth, with

the possible exception of immigrants from China where having more than one child is penalized.

Finally, we did not control for behavioural risk factors and maternal morbidity during pregnancy

because they are conceptualized as mediators in the relationship between socioeconomic factors

and birth outcomes.1,57,58

Neighbourhood context was assessed at the time of delivery but some of the mothers may have

been exposed to more than one neighbourhood within the study period, the probability being

higher for those mothers who took longer to have their first Canadian-born infant. A lack of

information about the residential trajectory of the mothers prevented us from assessing the extent

of this bias. Residential mobility is a complex phenomenon that may be influenced by individual

94

and neighbourhood characteristics and class relations, and vary by ethnic group, nativity status

and length of residence. 59-61 It is unlikely, however, that residential mobility would have

introduced serious bias since our study population was restricted to recent immigrants, with short

length of stay. Moreover, a recent study did not find substantial differences between longitudinal

and cross-sectional estimates of neighbourhood effects on children well-being, 60 suggesting that

most families moved between neighbourhoods of the same socioeconomic type, which is

consistent with the research of South and colleagues 61 on inter-neighbourhood socioeconomic

mobility.

We used four independent indices of neighbourhood context, based on census data that

essentially reflect aggregated characteristics of the population, but did not use measures based on

other data sources that may have provided information about other aspects of the residential

milieu. In Canada, census tract boundaries have been found to correspond well to those of

‘natural’ neighbourhoods. 62 Several Canadian studies have found significant area-level effects

using this geographic unit of analysis with a broad array of outcomes, 38,63-66 including birth

outcomes. 28,67

We did not expect higher birthweight among refugees, compared with non-refugee immigrants,

since they usually emigrate from high-stressed environments, which could lead to adverse birth

outcomes. The birthweight advantage of refugees was reduced in the adjusted models but it

remained significant, implying a role of unmeasured factors. Studies comparing obstetric

outcomes of refugees and asylum-seekers from Somalia and Kosovo-Albania with those of UK-

born and US-born White women did not find significant differences, 41,68,69 probably due to small

sample sizes.

95

We also found that the trend for birth weight increased with length of residence, even after

adjustment. Although the reasons for this finding in Ontario are not clear, previous research

suggests that birth outcomes may either improve or deteriorate with length of residence among

first generation immigrants, depending on the migrant group and/or the receiving environment.

For instance, the risk of preterm birth and low-birthweight increased with length of residence

among Mexican migrants to the U.S. 29,44 and Asians and Pacific Islanders in Sweden but

decreased among Finns, 45 while other migrant groups were unaffected.

Our finding regarding the absence of significant neighbourhood effects on birthweight among

immigrants is not surprising given the inconsistent associations with measures of socioeconomic

position found among diversely ethnic immigrant groups to the United States. 19,21,56,70,71 The

finding of important country-level effects is consistent with a literature reporting wide variability

in pregnancy outcomes by world region, country and ethnic groups.26,71 Although our findings

are consistent with a recent study in which the association between residential segregation and

low birthweight disappeared after controlling for nativity among immigrant Black women to

New York City,12 generalization to North American cities with little ethnic and nativity diversity

may be limited.

Figure 4.2 displays a simplified causal diagram conceptualizing the interplay of country of birth

and neighbourhood deprivation on birthweight. (Appendix 3.D for a more complex causal

diagram) Self-selection of immigrants to neighbourhoods according to their countries of origin

creates a modest spurious association between neighbourhood deprivation and birthweight,

which is removed by controlling by country of birth. The corollary is that if we are interested in

the independent association between neighbourhood deprivation and birthweight then we should

control for country of birth. But if our analytic goal is to assess the total association between

96

immigration/country of birth and birthweight then we should not control for neighbourhood

context. It is important to remember that the diagram conceptualizes what happens among recent

immigrants, but it may not be valid for immigrants with longer stay in Canada.

Figure 4.2: Confounding by self-selection of recent immigrants to neighbourhoods

Our findings are also consistent with three-level studies that showed significant reductions in the

amount of variability attributable to the neighbourhood context, after taking account of an

additional context such as the family or household. 72-74 Such evidence suggests that two-level

studies of neighbourhood effects may overestimate the contribution of the residential

environment if they disregard other contexts potentially relevant to the population and outcome

under study.

Interventions and policy recommendations at the neighbourhood level should be cautious if

based on studies that did not control for other meaningful contexts for the study population. We

did not find evidence that neighbourhoods matter for immigrants’ offsprings’ birthweight but this

cannot be generalized to other outcomes without further empirical research. These findings could

help to direct prenatal and even pre-conception programming towards recent immigrant women

97

from higher-risk countries of origin. While neighbourhood may not be a major exposure for low

birthweight in urban Ontario, it can provide a vehicle to reach out to women that are at particular

risk upon entry into Canada through facilitating local access to culturally sensitive prenatal care

and translation services.

Credits

This chapter represents a prepublication version of the following article:

Urquia ML, Frank JW, Glazier RH, Moineddin R, Matheson FI, Gagnon AJ. Neighborhood

context and infant birthweight among recent immigrant mothers: a multilevel analysis. Am J

Public Health. 2009 Feb;99(2):285-93.

98

4.5. References



(2) Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, Carson R. Choosing area based

socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning:

The Public Health Disparities Geocoding Project (US). J Epidemiol Community Health. 2003;57(3):186-

199.

(3) Spencer N, Bambang S, Logan S, Gill L. Socioeconomic status and birth weight: comparison of an area-

based measure with the Registrar General's social class. J Epidemiol Community Health. 1999;53(8):495-

498.

(4) Gudmundsson S, Bjorgvinsdottir L, Molin J, Gunnarsson G, Marsal K. Socioeconomic status and perinatal

outcome according to residence area in the city of Malmo. Acta Obstet Gynecol Scand. 1997;76(4):318-

323.

(5) Luo ZC, Wilkins R, Kramer MS. Effect of neighbourhood income and maternal education on birth

outcomes: a population-based study. CMAJ. 2006;174(10):1415-1420.

(6) Joseph KS, Liston RM, Dodds L, Dahlgren L, Allen AC. Socioeconomic status and perinatal outcomes in a

setting with universal access to essential health care services. CMAJ. 2007;177(6):583-590.

(7) Bell JF, Zimmerman FJ, Almgren GR, Mayer JD, Huebner CE. Birth outcomes among urban African-

American women: a multilevel analysis of the role of racial residential segregation. Soc Sci Med.

2006;63(12):3030-3045.

99

(8) Buka SL, Brennan RT, Rich-Edwards JW, Raudenbush SW, Earls F. Neighbourhood support and the birth

weight of urban infants. Am J Epidemiol. 2003;157(1):1-8.

(9) Dibben C, Sigala M, Macfarlane A. Area deprivation, individual factors and low birth weight in England: is

there evidence of an "area effect"? J Epidemiol Community Health. 2006;60(12):1053-1059.

(10) Farley TA, Mason K, Rice J, Habel JD, Scribner R, Cohen DA. The relationship between the

neighbourhood environment and adverse birth outcomes. Paediatr Perinat Epidemiol. 2006;20(3):188-200.

(11) Grady SC. Racial disparities in low birthweight and the contribution of residential segregation: a multilevel

analysis. Soc Sci Med. 2006;63(12):3013-3029.

(12) Grady SC, McLafferty S. Segregation, Nativity, and Health: Reproductive Health Inequalities for

Immigrant and Native-Born Black Women in New York City. Urban Geography. 2007;28(4):377-397.

(13) Jarvelin MR, Elliott P, Kleinschmidt I et al. Ecological and individual predictors of birthweight in a

northern Finland birth cohort 1986. Paediatr Perinat Epidemiol. 1997;11(3):298-312.

(14) Morenoff JD. Neighbourhood mechanisms and the spatial dynamics of birth weight. AJS. 2003;108(5):976-

1017.

(15) O'Campo P, Xue X, Wang MC, Caughy M. Neighbourhood risk factors for low birthweight in Baltimore: a

multilevel analysis. Am J Public Health. 1997;87(7):1113-1118.

(16) Rauh VA, Andrews HF, Garfinkel RS. The contribution of maternal age to racial disparities in birthweight:

a multilevel perspective. Am J Public Health. 2001;91(11):1815-1824.

100

(17) Subramanian SV, Chen JT, Rehkopf DH, Waterman PD, Krieger N. Comparing individual- and area-based

socioeconomic measures for the surveillance of health disparities: A multilevel analysis of Massachusetts

births, 1989-1991. Am J Epidemiol. 2006;164(9):823-834.

(18) O'Campo P, Burke JG, Culhane J et al. Neighbourhood deprivation and preterm birth among non-Hispanic

Black and White women in eight geographic areas in the United States. Am J Epidemiol. 2008;167(2):155-

163.



(20) Peak C, Weeks JR. Does community context influence reproductive outcomes of Mexican origin women in

San Diego, California? J Immigr Health. 2002;4(3):125-136.



(22) O'Campo P. Invited commentary: Advancing theory and methods for multilevel models of residential

neighbourhoods and health. Am J Epidemiol. 2003;157(1):9-13.

(23) Davey Smith G, Lynch J. Life course approaches to socioeconomic differentials in health. In: Kuh D,

Shlomo B, eds. A Life Course Approach to Chronic Disease Epidemiology. New York: Oxford University

Press, 2004: 77-115.





101


Int J Gynaecol Obstet. 2003;83(3):257-265.

(26) United Nations Children's Fund and World Health Organization. Low Birthweight: Country, Regional and

Global Estimates. 2004. New York, UNICEF.







(30) Statistics Canada. 1996 Census Dictionary - Final Edition. [Statistics Canada]. 1999. Available at:

http://www.statcan.ca/bsolc/english/bsolc?catno=92-351-U&CHROPG=1. Accessed February 2, 2008.

(31) Statistics Canada. Standard Geographical Classification (SGC) 1996. [Statistics Canada]. 2008. Available

at: http://www.statcan.ca/english/Subjects/Standard/sgc/1996/1996-sgc-index.htm. Accessed January 25,

2008.


1241.

(33) World Bank. Classification of Economies by Income and Region, 2000. [http://www worldbank org]. 2007.

(34) Diez Roux AV. Residential environments and cardiovascular risk. J Urban Health. 2003;80(4):569-589.

102

(35) Hill TD, Ross CE, Angel RJ. Neighbourhood disorder, psychophysiological distress, and health. J Health

Soc Behav. 2005;46(2):170-186.

(36) Pampalon R, Raymond G. A deprivation index for health and welfare planning in Quebec. Chronic Dis

Can. 2000;21(3):104-113.

(37) Sampson RJ, Raudenbush SW, Earls F. Neighbourhoods and violent crime: a multilevel study of collective

efficacy. Science. 1997;277(5328):918-924.

(38) Matheson FI, Moineddin R, Dunn JR, Creatore MI, Gozdyra P, Glazier RH. Urban neighbourhoods,

chronic stress, gender and depression. Soc Sci Med. 2006;63(10):2604-2616.

(39) World Health Organization. International Classification of Diseases, Injuries and Causes of Death. 9th

revision. Geneva: The Organization; 1979.

(40) Kuvacic I, Skrablin S, Hodzic D, Milkovic G. Possible influence of expatriation on perinatal outcome. Acta

Obstetricia et Gynecologica Scandinavica. 1996;75(4):367-371.

(41) Yoong W, Kolhe S, Karoshi M, Ullah M, Nauta M. The obstetric performance of United Kingdom asylum

seekers from Somalia: a case-control study and literature review. International Journal of Fertility &

Womens Medicine. 2005;50(4):175-179.



(43) Rumbaut R, Weeks J. Unraveling a public health enigma: Why do immigrants experience superior perinatal

health outcomes. Research in the Sociology of Health Care. 1996;13B:337-391.

103






(46) Beretvas SN, Meyers JL, Rodriguez RA. The cross-classified multilevel measurement model: an

explanation and demonstration. J Appl Meas. 2005;6(3):322-341.

(47) Raudenbush SW, Bryk AS. Hierarchical Linear Models. Applications and Data Analysis Methods. second

ed. Thousand Oaks, CA: Sage Publications; 2002.


2002.

(49) Maas CJM, Hox JJ. Sufficient Sample Sizes for Mutilevel Modeling. Methodology. 2005;1(3):86-92.

(50) Snijders T, Bosker RJ. Multilevel Analysis. An Introduction to basic and advanced multilevel modeling.

London: Sage Publications; 1999.

(51) Buekens P, Masuy-Stroobant G, Delvaux T. High birthweights among infants of north African immigrants

in Belgium. Am J Pub Health. 1998;88(5):808-811.

(52) Delvaux T, Buekens P, Thoumsin H, Dramaix M, Collette J. Cord C-peptide and insulin-like growth factor-

I, birth weight, and placenta weight among North African and Belgian neonates. American Journal of

Obstetrics & Gynecology. 2003;189(6):1779-1784.

104

(53) Guendelman S, Buekens P, Blondel B, Kaminski M, Notzon FC, Masuy-Stroobant G. Birth outcomes of

immigrant women in the United States, France, and Belgium. Matern Child Health J. 1999;3(4):177-187.

(54) Vahratian A, Buekens P, Delvaux T et al. Birthweight differences among infants of North African

immigrants and Belgians in Belgium. European Journal of Public Health. 2004;14(4):381-383.

(55) Gayral-Taminh M, Arnaud C, Parant O et al. [Pregnancy and labor of women born in Maghreb and Black

Africa followed to delivery at the Maternity Hospital of Toulouse]. [French]. Journal de Gynecologie,

Obstetrique et Biologie de la Reproduction. 1999;28(5):462-471.



(57) Kramer MS. Determinants of low birth weight: methodological assessment and meta-analysis. Bull World

Health Organ. 1987;65(5):663-737.

(58) Kramer MS, Goulet L, Lydon J et al. Socio-economic disparities in preterm birth: causal pathways and

mechanisms. Paediatr Perinat Epidemiol. 2001;15 Suppl 2:104-123.

(59) Clark WAV, Ledwith V. How much does income matter in neighbourhood choice? Population Research

and Policy Review. 2007;26:145-161.

(60) Jackson MI, Mare RD. Cross-Sectional and Longitudinal Measurements of Neighbourhood Experience and

Their Effects on Children. Social Science Research. 2007;36:590-610.

(61) South SJ, Crowder K, Chavez E. Exiting and entering high-poverty neighbourhoods: Latinos, Blacks and

Anglos compared. Social Forces. 2005;82(2):873-900.

105

(62) Ross NA, Tremblay SS, Graham K. Neighbourhood influences on health in Montreal, Canada. Soc Sci

Med. 2004;59(7):1485-1494.

(63) Glazier RH, Creatore MI, Gozdyra P et al. Geographic methods for understanding and responding to

disparities in mammography use in Toronto, Canada. J Gen Intern Med. 2004;19(9):952-961.

(64) James PD, Wilkins R, Detsky AS, Tugwell P, Manuel DG. Avoidable mortality by neighbourhood income

in Canada: 25 years after the establishment of universal health insurance. J Epidemiol Community Health.

2007;61(4):287-296.

(65) Macpherson AK, Macarthur C, To TM, Chipman ML, Wright JG, Parkin PC. Economic disparity in

bicycle helmet use by children six years after the introduction of legislation. Inj Prev. 2006;12(4):231-235.

(66) McLaren L, Gauvin L. Neighbourhood level versus individual level correlates of women's body

dissatisfaction: toward a multilevel understanding of the role of affluence. J Epidemiol Community Health.

2002;56(3):193-199.

(67) Wilkins R, Sherman GJ, Best PA. Birth outcomes and infant mortality by income in urban Canada, 1986.

Health Rep. 1991;3(1):7-31.

(68) Johnson EB, Reed SD, Hitti J, Batra M. Increased risk of adverse pregnancy outcome among Somali

immigrants in Washington state. American Journal of Obstetrics & Gynecology. 2005;193(2):475-482.

(69) Yoong W, Wagley A, Fong C, Chukwuma C, Nauta M. Obstetric performance of ethnic Kosovo Albanian

asylum seekers in London: a case-control study. Journal of Obstetrics & Gynaecology. 2004;24(5):510-

512.

106




pregnancy outcomes in two dissimilar immigrant groups in the United States. J Pediatr. 2006;148(3):341-

346.

(72) Boyle MH, Lipman EL. Do places matter? Socioeconomic disadvantage and behavioral problems of

children in Canada. J Consult Clin Psychol. 2002;70(2):378-389.

(73) Chandola T, Clarke P, Wiggins RD, Bartley M. Who you live with and where you live: setting the context

for health using multiple membership multilevel models. J Epidemiol Community Health. 2005;59(2):170-

175.

(74) Sacker A, Wiggins RD, Bartley M. Time and place: putting individual health into context. A multilevel

analysis of the British household panel survey, 1991-2001. Health Place. 2006;12(3):279-290.

107

Chapter 5 The Differential Deterioration of Preterm Birth among urban

Immigrants by Neighbourhood Deprivation

Abstract

Socioeconomic gradients in birth outcomes seem to be ubiquitous but immigrants represent one

major exception to this pattern. Chapter 4 showed that neighbourhood context had no significant

influence on birthweight of recent immigrants, after adjusting for maternal country of birth.

However, those results are only valid among recent immigrants. The purpose of this chapter is to

examine the simultaneous contribution of the maternal country of birth and the neighbourhood

context on the preterm delivery of immigrants whose time from arrival in Canada to delivery

varied from 1 to 20 years. Using hierarchical models we examined whether neighbourhood

deprivation gradients in preterm birth were modified by duration of residence in Ontario urban

areas. There were no visible deprivation gradients among immigrants before 15 years of

residence. After 14 years, gradients for immigrants approached the gradients observed among

non-immigrants. Evaluation of neighbourhood influences among immigrants should pay close

attention to duration of residence.

108

5.1. Introduction

Socioeconomic gradients in birth outcomes are well documented. 1-4 Associations between low

SES and adverse birth outcomes can be observed through a wide range of measures, such as

individual income, education, occupation, or neighbourhood material deprivation and racial

segregation. However, immigrants represent one major exception to this pattern.5-12 Researchers

have elaborated some potential explanations to make sense of the inconsistent and even reversed

gradients found in various immigrant populations, such as Asian, Black, and Hispanic groups.

These include selective migration, low variability in socioeconomic status among immigrants,

and group-level attributes such as a protective cultural orientation.5,11,13

It is also possible that the lack of consistency of the socioeconomic gradients in birth outcomes

of migrants is affected by confounding by duration of residence in the receiving country. Recent

immigrants are more likely to be misclassified with respect to socioeconomic measures because

they experience a mismatch between their educational credentials and their initial insertion in the

labour market, and tend to settle in low-rental dwellings concentrated in poor

neighbourhoods,12,14 thus flattening the gradients. In contrast, measures of socioeconomic

position can be expected to be more consistent among long-term immigrants that have gone

through the adaptation process to the new physical and social environment and have reached a

relatively stable position in the new society. In addition, time would be needed for social

exposures, including neighbourhood deprivation, to exert its effects on health.15

We hypothesized that material deprivation gradients would tend to be more consistent with

increasing length of residence and examined whether the relation between neighbourhood

109

deprivation and preterm birth among immigrants to urban Ontario was modified by duration of

residence. We also compared immigrant gradients with those observed in the general population.

5.2. Methods

We extracted from the Discharge Abstract Database 397,470 singleton live births born to

mothers living in any of the 11 Ontario Census Metropolitan Areas16 at the time of delivery,

between April 1, 2002 and March 31, 2007, and with complete information on covariates. The

Landed Immigrant Data System (LIDS), which is the official immigration registry compiled by

Citizenship and Immigration Canada (CIC), was used to obtain sociodemographic and

immigration information of all legal immigrants that attained their permanent residence within

1995-2000, although some of them had temporary residence before, which is accounted for in

our measure of duration of residence defined as days from arrival to delivery. These data were

merged with small-area data (census tracts as neighbourhoods) from the 2001 Canadian census,

from which we constructed a material deprivation score17 with mean zero and standard deviation

one.

We first performed stratified analyses by duration groups, using a two-level model for non-

immigrants with births nested within neighbourhoods and cross-classified random effects models

(CCREM) for immigrants in order to account for the clustering within maternal countries of birth

and neighbourhoods. Details of this approach were given in Chapter 4.18 We calculated adjusted

odds ratios for the effect of one standard deviation of neighbourhood deprivation change on

preterm birth. We also used a 10% change in the population living below the Statistics Canada

low-income cut-off,19 for easier interpretation. Finally, we plotted predicted probabilities of

110

preterm birth based on a CCREM with all immigrants including an interaction term between

neighbourhood deprivation tertiles and duration of residence (p=0.0389).

5.3. Results

Table 5.1: Adjusted odds ratios (and 95% confidence intervals) of one Standard Deviation increase in the

material deprivation index and 10% increase in the population living below the low-income cut-off on

preterm birth, by immigrant status and duration of residence, urban Ontario, 2002-2007.

Births Coun

tries

Cens

us-

tracts

Country-level

Variance (SE)

Census-tract level

Variance (SE)

AOR per one SD

material

deprivation

(95% CI)

AOR per 10%

increase in

population below

LICO

(95% CI)

Non-

immigrants a 314,237 1 1801 0.019*** (0.004) 1.12 (1.10-1.14) 1.09 (1.07-1.11)

Immigrants by

duration b

<5 years

14,555 148 1472 0.125*

(0.067) 0.029 (0.061)

0.96 (0.89-1.03) 0.94 (0.88-1.00)

5-9 years 32,539 162 1649 0.101**

(0.040) 0.015 (0.019)

1.03 (0.99-1.08) 1.02 (0.98-1.07)

10-14 years

23,827 161 1622 0.045* (0.023) 0.009 (0.022) 1.01 (0.96-1.06) 1.01 (0.96-1.05)

≥ 15 years 12,312 133 1523 0.008 (0.022) 0.125** (0.054)

1.09 (1.02-1.17) 1.06 (1.00-1.13)

a based on a two-level model adjusted for infant sex, maternal age and parity. b based on a cross-classified model adjusted for infant sex, maternal age, parity, immigrant class, language knowledge, high school

graduation, maternal world region of birth, and unmarried status.

* p<0.05 ; ** p<0.01 ; *** p<0.001 (p-values for variances are one-sided)

Immigrants approached the association between the neighbourhood measures and preterm birth

observed in the general population after 15 years of stay (Table 5.1). The clustering of births to

immigrants, assessed by the neighbourhood variance, only reached statistical significance among

those with 15 or more years of residence, even after adjustment. It is noteworthy that among

111

these long-term immigrants, the variance for country of birth was no longer statistically

significant; implying that for this group, the neighbourhood context was more relevant. In other

words, country of birth is a good predictor of preterm birth among immigrants up to 14 years of

stay in Ontario but not afterwards. Conversely, measures of neighbourhood context only can

predict preterm birth among immigrants with 15 or more years of residence.

Figure 5.1. Predicted probabilities of preterm birth (2002-2007) by duration of residence and

neighbourhood deprivation tertiles among immigrants to urban Ontario

The risk of preterm birth increased with duration of residence in a cumulative dose-response

pattern across all neighbourhood deprivation tertiles. However, the deterioration of preterm birth

was attenuated (less steep slope) among those immigrants living in the least deprived

neighbourhoods at the time of delivery. The flat lines represent the levels (tertiles) of preterm

birth observed in the non-immigrant population living in low- and highly-deprived

neighbourhoods. Recent immigrants had lower preterm birth than non-immigrants living in low-

112

deprivation neighbourhoods, irrespective of the type of neighbourhood they lived in but they lost

their advantage over time, the sequence being consistent with the deprivation gradient.

5.4. Discussion

Consistent with the results obtained in Chapter 4, neighbourhood measures are not good

predictors of birth outcomes among recent immigrants. However, these new findings suggest that

they become more predictive with increasing duration of residence.

As our neighbourhood measures were based on the mothers’ residence at the time of delivery

and many may have moved since arrival, our findings cannot be interpreted as resulting from

cumulative exposure to neighbourhood deprivation levels. Rather, the type of neighbourhoods in

which mothers lived in at delivery is conceptualized here as a marker of the socioeconomic

position achieved by that time.

Unlike Chapter 4 that is restricted to recent immigrants, this Chapter also compares the gradients

found among immigrants with those found in the majority population. Although the levels of

preterm birth among immigrants did not ‘converge’ with time spent in urban Ontario towards the

average observed in the non-immigrant population, as discussed in Chapter 3, the socioeconomic

gradient in preterm birth (i.e., the distance between tertiles 1 an 3) among immigrants

approximated the gradient observed in the majority population after 14 years of stay

approximately, this being consistent with a pattern of convergence.

Duration of residence can be considered a key dimension in studies of immigrants’ health.

113

5.5. References

(1) Grady SC. Racial disparities in low birthweight and the contribution of residential segregation: a multilevel

analysis. Soc Sci Med. 2006;63(12):3013-3029.



(3) Krieger N, Chen JT, Waterman PD, Soobader MJ, Subramanian SV, Carson R. Choosing area based

socioeconomic measures to monitor social inequalities in low birth weight and childhood lead poisoning:

The Public Health Disparities Geocoding Project (US). J Epidemiol Community Health. 2003;57(3):186-

199.

(4) O'Campo P, Xue X, Wang MC, Caughy M. Neighbourhood risk factors for low birthweight in Baltimore: a

multilevel analysis. Am J Public Health. 1997;87(7):1113-1118.



(6) David RJ, Collins JW, Jr. Differing birth weight among infants of U.S.-born blacks, African-born blacks,

and U.S.-born whites.[see comment]. New England Journal of Medicine. 1997;337(17):1209-1214.





114

(9) Landale NS, Oropesa, Gorman BK. Immigration and infant health:birth outcomes of immigrant and native-

born women. Children of Immigrants: Health, Adjustment and Public Assistance. Washington, DC:

National Academy Press, 1999: 244-85.


pregnancy outcomes in two dissimilar immigrant groups in the United States. J Pediatr. 2006;148(3):341-

346.







109.

(14) Galarneau D, Morissette R. Immigrants: Settling for Less? Perspectives on Labor and Income. 2004;5(6):5-

16.

(15) O'Campo P. Invited commentary: Advancing theory and methods for multilevel models of residential

neighbourhoods and health. Am J Epidemiol. 2003;157(1):9-13.

(16) Statistics Canada. Standard Geographical Classification (SGC). Volume I. The Classification. [Statistics

Canada]. 2007. Accessed December 12, 2008.

115

(17) Matheson FI, Moineddin R, Dunn JR, Creatore MI, Gozdyra P, Glazier RH. Urban neighbourhoods,

chronic stress, gender and depression. Soc Sci Med. 2006;63(10):2604-2616.

(18) Urquia ML, Frank JW, Glazier RH, Moineddin R, Matheson FI, Gagnon AJ. Neighbourhood Context and

Infant Birthweight Among Recent Immigrant Mothers: A Multilevel Analysis. Am J Pub Health.

2009;99(2):1-9.

(19) Statistics Canada. Low Income Cut-offs. [Statistics Canada]. 1999. Accessed December 11, 2008.

116

Chapter 6 Discussion

6.1. Main Findings

The pieces of work comprising this thesis indicate that:

a) Ethnicity, country of origin, and time since migration are important predictors of birth

outcomes among immigrants.

b) Duration of residence is linearly associated with increases in low birth weight and preterm

birth, mainly driven by decreases in gestational age with prolonged stay in Canada.

c) The detrimental effects of long duration of residence on preterm birth are modestly attenuated,

but not prevented, among immigrants living in urban neighbourhoods characterized by low

material deprivation.

d) Neighbourhood material deprivation has little, if any, influence on birth outcomes of recent

immigrants but the influence of neighbourhood deprivation becomes visible with longer stay in

Canada. Maternal world region of origin constitutes a stronger predictor among recent

immigrants.

117

6.2. Strengths and Limitations

Although the strengths and limitations have been discussed in each paper, it is convenient to

briefly highlight here the main ones, considering all the foregoing chapters in the context of the

broader literature.

One of the common strengths is that all analyses are based on relatively large sample sizes, in

most cases population-based.

Although it is virtually impossible to control for all relevant covariates, due to data limitations

and ignorance about what the relevant confounders are, the 95% confidence intervals of the

effect estimates obtained from these analyses partially account for unknown sources of

heterogeneity, since hierarchical models were used to account for the clustering of births within

studies and subgroups within studies in the meta-analyses (Chapter 2) and within

neighbourhoods and maternal countries of birth in the analyses using Ontario data (Chapters 3-

5).

Regarding the potential confounders with available data, selection of covariates was based on

causal diagram theory,1,2 which improves the likelihood of obtaining unbiased effect estimates

by avoiding improper control. However, the helpfulness of the causal diagrams depends on our

knowledge of what the potential confounders are, which is not exhaustive in this area of

research.

The Ontario birth and immigration data spanned several years to allow estimation of the effects

of duration of residence over 20 years since arrival, which has not been done before for birth

outcomes. Moreover, with the exception of knowledge of official languages, which was self-

reported, the remaining immigration data were collected through the documentation provided by

118

the immigrants during the Canadian-entry application process and last updated at the port of

entry on the landing date.

One common limitation in our analyses, as well as in the literature, is the use of data for legal

international migrants only. Temporary, internal, and illegal migration are not captured by the

Ontario immigration data, and is rarely and inconsistently recorded in birth certificates

worldwide, which represent the main data source of the studies included in the meta-analyses.

Finally, there is a generalized lack of information on emigrants of the receiving countries, who

represent another unmeasured influence shaping the health outcomes of the ‘native’ populations

(i.e., comparison groups).3

A second common limitation found in the literature is the heterogeneity in the definition of what

an immigrant is and its measurement. Different migrant labels have been used in the reviewed

literature, such as “foreign-born”, not always specifying the country of birth/origin, and

nationality.4 Although the Ontario immigrant data constitute an excellent source to identify

immigrants, the dataset used started with migrants landed in 1985. Therefore, immigrants landed

before 1985 were misclassified as non-immigrants thus slightly biasing the results towards the

null; “slightly” because immigrant women who arrived before 1985 were somewhat less likely to

deliver babies at least 17 years after arrival (2002-2007) and the proportion of babies born to

immigrant women decreases dramatically with such long duration of residence; “towards the

null” because duration of residence is associated with increased risk of low birthweight and

preterm birth and high-risk immigrant women with long duration of residence were counted as

part of the non-immigrant population, thus narrowing the differences between long-term

immigrants and non-immigrants.

119

A third common limitation is that although the outcomes analyzed in this thesis have been

widely used for surveillance, they are imperfect outcomes for etiologic research.5,6 Birthweight

differences by migrant subgroups generally focus on one parameter of the birthweight

distribution (i.e., mean birthweight), which may not be the most relevant for perinatal health.6

Despite the strong associations found between low birthweight and infant mortality, low

birthweight alone is ill-suited for etiologic studies, since it cannot discriminate between those

infants who are small because of being born preterm and those with growth restriction. In terms

of perinatal knowledge, the models of low birthweight do not provide significant additional

information beyond the information provided by the models of preterm birth and small for

gestational age (Chapter 3). However, the models of low birthweight were kept to allow

comparisons with the broader literature, which has mainly focused on that outcome. In addition,

and despite its acceptance, the 2,500 grams cut-off has been criticized for being arbitrary.6 This

same criticism can be applied to the 37 weeks cut-off defining preterm birth.7 Moreover, it has

been suggested that gestational age is in fact follow up time and should be better considered as

such in causal models rather than as an endpoint on its own.5 Finally, the measurement of intra-

uterine fetal growth still remains a challenge.8 Small for gestational age represents a proxy based

on cross sectional estimates (sex-specific birthweight percentiles by week of gestation) at birth

but does not accurately reflect the longitudinal process of fetal development over the gestation

period,6,8 which may vary across subpopulations (e.g., ethnic groups). The use of a Canadian-

based standard for small for gestational age may not be appropriate to assess disparities by

migration status, since differences between ethnic groups have been suggested to be more

physiological than pathological.9 Caution is therefore advised when interpreting the reported

differences in birthweight-related outcomes between migrant subgroups. Low birthweight,

preterm birth and small for gestational age are intermediate outcomes in perinatology that have

120

been associated with perinatal mortality, serious neonatal morbidity, and childhood disabilities.

Although perhaps more relevant, these endpoints were not chosen as the main outcomes of this

thesis because of data quality concerns and measurement issues. In addition, these are rare

outcomes that may have reduced the power to detect statistically significant differences for

various comparison groups, given our sample sizes.

Another limitation is that our main analyses assessing the influence of duration of residence on

birth outcomes (Chapters 3 and 5) were based on cross-sectional estimates. Although we

conducted sensitivity analyses on a cohort of immigrants who arrived during 1995-1998 and by

subsequent 4-year cohorts, yielding results consistent with the main analyses, we choose not to

report the cohort estimates as the main analyses because this approach suffered from other

weaknesses. These comprise: our inability to measure some relevant outcomes (very preterm

birth and small for gestational age); the use of different methods for ascertaining preterm birth

(ICD-9 codes from fiscal 1988/1989 to 2001/2002 and gestational age in weeks from 2002/2003

to 2006/2007); potential confounding by secular increases in preterm birth due to obstetric

practice and other unmeasured maternal characteristics;10,11 and the lack of information on parity

prior to fiscal 2002/2003.

Another limitation is that we could not provide satisfactory explanations of why preterm birth

increases with duration of residence among immigrants. The testing of some hypothetical

explanations as found in the literature, such as changes in health behaviours and job-related

exposures, requires data that were not available in the datasets used for this thesis. In particular,

the lack of information on maternal smoking and pre-pregnancy weight prevented the testing of

mediation by these factors when assessing the association between duration of residence and

preterm birth.

121

Generalization of findings is also difficult. Given the diversity of migrant groups, defined by

their country, region, and city of origin, ethnicity, religion, social class background, and of the

receiving environments, defined by their country, region, city, labour market and social

dynamics, it is virtually impossible to generalize the findings of this thesis and predict the

outcomes of immigrants in particular situations. Moreover, our findings may not be generalizable

to countries where access to health care is not universal (e.g., United Sates), thus introducing

another potential source of disparities. However, this thesis provides some insights that may help

health practitioners and policymakers avoid misleading approaches when focusing on immigrant

groups, and may help researchers design sounder studies to advance knowledge in this field.

6.3. Implications for practice and future research

The first insight from this work is that attention must be paid to the definition of migrant groups

and choice of comparison groups, when drawing conclusions about disparities by migrant status.

As shown in the literature review, a specific migrant group (e.g., Sub-Saharan Africans in the

US) can be easily labelled as ‘protected’ or ‘at high-risk’ merely by changing the referent group

from ‘US-born Blacks’ to ‘US-born Whites’. Thus, arbitrary manipulation of comparison groups

may result in contrasting conclusions. Moreover, as shown in Chapters 3 and 5, disparities by

migrant status are modified by duration of residence. Recent immigrants compare favourably to

non-immigrants but the opposite is true for long-term immigrants. Such differences may be

bigger when socioeconomic gradients are also considered. This leads us to a related issue.

The second insight from this work is that health studies of immigrants defined as a single

category are not informative. All chapters have shown substantial heterogeneity among

immigrants according to their world region of origin and duration of residence in the receiving

122

country, with specific migrant subgroups experiencing risks below and above the average, and

below and above the native-born. Grouping together all immigrants not only masks these

important heterogeneities, but also generates misleading statistics to inform policy and

interventions directed to particular migrant groups. Differences in birth outcomes between

immigrants from various regions of the world may be greater than differences found with

established predictors, such as maternal education, marital status, parity, or some pregnancy

complications. Duration of residence is especially important because immigrants can be seen as a

cohort exposed to a new environment starting on arrival. Although duration of residence can be

conceptualized as a marker for other exposures (e.g., acculturation), it remains of interest in itself

because it measures time of exposure to the new environment (i.e., follow-up time). Therefore,

both country of birth/origin and duration of residence have been found to be key predictors of

birth outcomes among immigrants. These two predictors should be considered in future studies

and also for surveillance, whenever feasible. Prenatal and women’s health programs would also

benefit from separate clinical management of those migrant subgroups at higher risk of adverse

birth outcomes: long-term immigrants (particularly those living in poor neighbourhoods), and

immigrants from South Asia, Sub-Saharan Africa, and Latin America and Caribbean.

The association between duration of residence and preterm birth might have important

implications for public health. Preterm birth has been considered to be a determinant of other

adverse health outcomes in the perinatal period (e.g., acute neonatal illness, perinatal

mortality)12, and long-term consequences.13 Preterm survivors are at higher risk of

neurodevelopmental disabilities that in turn make them more prone to develop language

disorders, learning disabilities, attention deficit-hyperactivity disorder, and behavioural

problems. Preterm infants have lower intelligence quotients and academic achievement scores,

experience greater difficulties at school, and require significantly more educational assistance

123

than children who were born at term. They also have an increased risk of re-hospitalization

during the first few years of life, increased use of outpatient care, and chronic health disorders in

adulthood.13,14 Thus, the long-term consequences of preterm birth have been estimated to

produce a substantial financial burden on the health care system15-17 and are likely to spill over

into other sectors of society. The existence of these sequelae provides a strong rationale for the

prevention of preterm birth. However, the increasing trends in preterm birth in recent years in the

general population of North America, mostly driven by obstetric interventions at 34-36 weeks of

gestation (medically indicated preterm delivery), were not accompanied by increases in infant

mortality (indeed infant mortality decreased) or in certain neurodevelopmental disabilities, such

as cerebral palsy.5,10,18 Whether these secular trends observed in the general population apply to

immigrants to Ontario is something to be confirmed by empirical research. The investigation of

these issues is likely to advance knowledge on the links between migration and perinatal health

and beyond. The prevention of the deterioration of gestational age among immigrants over time

spent in Canada may represent a worthy public health priority, if further research can better

clarify the causal pathways involved, and identify candidate interventions to be successfully

applied to different migrant groups. To date, most interventions to reduce preterm birth have

yielded disappointing results.18,19 Success was achieved in reducing the negative consequences of

preterm birth (tertiary prevention) due to intensive and improved obstetric and neonatal care

rather than preterm birth itself.18 While some risk factors affecting migrant women are not

modifiable (e.g., genetics, maternal height, early-life exposures), migrant women could benefit

from primary and secondary prevention of preterm birth (targeting health behaviours,

overweight/obesity, diet, removing barriers to women’s health and prenatal services, community

outreach for identification of women at risk, control of risk factors during pregnancy), but these

strategies are not receiving as much attention as tertiary interventions.18

124

6.4. Unanswered questions and future research

Our understanding of the relation between reproductive health and international migration

remains poor. Here, I pose a few questions for which answers may help advance knowledge in

this area and briefly discuss some related study design issues.

6.4.1. Why is duration of residence associated with preterm birth?

The first question is what the mechanisms are through which time of exposure to urban Ontario

leads to higher risk of preterm birth. Connecting the dots between time spent in Canada after

migration and preterm birth represents a challenge, since several mediators may be operating

simultaneously forming one or more complex bio-psycho-social causal chains. Based on the

literature, it can be hypothesized that the following types of mediators are likely to be involved:

social environment, health behaviours, maternal illnesses before gestation, maternal

complications during pregnancy, and obstetric practices. First, the social environment is not

really a mediator but rather the broad context of acculturation, the cumulative influence of which

was proxied for by duration of residence after migration. The conceptualization and

decomposition of the social environment into measurable characteristics may help identify

specific stressors operating after migration. Such stressors may be influences operating in the

residential environment, labour market and workplace effects, as well as gender, ethno-cultural,

and psychological problems resulting from parental insertion in the Canadian society.20-24

Second, the social environment may exert its effects on health by modifying individual health

behaviours.25-28 Changes in maternal health behaviours and risk factors for preterm birth are

likely to be at least partially responsible for the increase in preterm birth risk experienced by

migrants with time spent in Canada. Indeed, deterioration of health behaviours and risk factors

125

after migration is supported by non-negligible evidence in the U.S.26,29-31, the U.K.,27 and also in

Canada.20,25,32,33

Another unexplored aspect of acculturation to Western societies may involve increased

medicalization with time spent in the new society. Such medicalization may be translated in a

shortening of the time from conception to first prenatal visit, increased compliance and enhanced

prenatal surveillance, and preference for elective cesarean delivery. When compared with

menstrual dates, early ultrasound based dating results in a left shift in the gestational age

distribution. If immigrants with longer duration of residence were more medicalized and enter

antenatal care earlier, their gestational age estimate might has been more likely to be based on

early ultrasound than among less medicalized immigrant women who had their first prenatal visit

at a more advanced stage of their pregnancies. Even assuming complete polarization of early

ultrasound use in our study population, this could not completely explain our results, since the

bias resulting from using ultrasound dating only versus the last menstrual period only was

estimated to be no more than a 10-20% increase in the risk of preterm birth,34 and we observed a

14%, 30%, and 49% increased odds associated with 5, 10, and 15 years of residence,

respectively. Another potential pathway related to increased medicalization may be increased

preference for elective cesarean section, which may shorten gestation length for some women. It

is also possible that use of assisted reproductive technologies (ARTs) have become more

frequent among the more acculturated migrant couples. Such a hypothetical pathway is plausible

for the following reasons: recent immigrants are on average less likely to have multiple

gestations and make use of ARTs than the general population; ARTs use is associated with

higher incomes35 and immigrants in Canada improve their income with time spent in Canada,

and use of ARTs has been associated with increased preterm birth even among singleton

pregnancies.36 Third, pre-existing maternal conditions and complications arising during

126

pregnancy are strong predictors of preterm birth and other adverse birth outcomes.37,38 Some of

these maternal conditions (e.g., hypertension, diabetes) are the result of underlying risk factors

such as lack of physical activity, unhealthy diet, overweight/obesity, and tobacco smoking,

which in turn may have been acquired by a number of women at some point after migration. The

plausibility of this pathway for our study population is supported by one Ontario-based study

reporting that the lower risk of maternal placental syndrome (assessed by a diagnosis of pre-

eclampsia or eclampsia, placental abruption or placental infarction) among newly arrived

immigrants came close to the level observed in the majority population (Canadian-born and

immigrants with more than 5 years of residence) in five years, presumably due to adoption of a

Western-diet and lifestyle.33 However, we have observed opposite trends in other maternal

conditions leading to preterm birth, such as genito-urinary infections, which may cancel out the

explanatory power of maternal conditions positively associated with time since migration.

Specific studies should be designed to weight the contribution of diverse maternal conditions on

the observed association between time since migration and preterm birth. Fourth, the existence of

maternal medical conditions affecting either the mother or the fetus constitutes one major reason

for medical interventions leading to preterm deliveries, which account for about one third of all

preterm births,37 either by induction of labour or by pre-labour caesarean section. Although

studies assessing the association between maternal obesity/overweight and preterm birth have

produced mixed results,14,39-43 there is increasing evidence supporting the existence of a link

between maternal overweight/obesity and pregnancy complications, such as gestational diabetes

and pre-eclampsia.44,45 High maternal pre-pregnancy weight and weight gain during pregnancy

have also been found to be associated with caesarean sections.11,46 These associations between

overweight/obesity and immediate predictors of preterm birth make the hypothesized pathway

“migrant cumulative exposure to Canada increase in body weight indicated preterm birth”

127

a plausible one. Specific studies would be needed to test whether and to what extent maternal

weight gain with time spent in Canada was responsible for increases in preterm birth among

immigrants, and if such hypothetical association varied by preterm birth types. The challenge,

however, is to obtain data on maternal pre-pregnancy weight, since maternal anthropometric

measures are not available in routinely collected administrative data. Survey data (various

Canadian Community Health Survey waves) linkable to administrative hospital data contain self-

reported weight and height, but these measures were obtained at the interview, and therefore not

around conception among the subset reporting having a baby in the 5 years preceding the

interview. However, these survey data can be used as secondary evidence to assess prevalence in

anthropometric, lifestyle, and psychosocial risk factors by time since migration.

Preterm births not caused by medical intervention are categorized as spontaneous, which in turn

subdivide into spontaneous preterm labour with intact membranes, and preterm premature

rupture of membranes (PPROM), irrespective of the mode of delivery (i.e., vaginal or caesarean).

Risk factors (both distal and proximal) for preterm birth were reported to differ between these

subtypes.37,47 One such factor is ethnic group, which is closely related to world region of origin

among migrants. For instance, spontaneous preterm birth is most commonly caused by preterm

labour among White women, but by PPROM among Black women in the U.S.19,37 This example

suggests that pathways linking immigrants’ duration of residence and preterm birth may also be

ethnic-specific. One potential strategy would be to restrict future studies to specific migrant

groups to avoid residual confounding. The task of connecting the dots between time spent after

migration and preterm birth may become even more complex and challenging if there is effect

modification by factors other than ethnic group.

128

Finally, it is also likely that duration of residence in Canada is an effect modifier of some risk

factors on its own. We have already found that the effects of duration of residence interacted

with neighbourhood deprivation (Chapter 5). It can be expected that some of the risk factors

hypothesized to be in the causal pathway “time since migration preterm birth” may be

present or absent at certain times or their effect may vary over time spent in Canada. For

example, cultural and language barriers may be more critical among recent immigrants but no

longer be a source of stress some years after arrival. In the same vein, exposure to jobs below

maternal educational and skill level is most likely to occur among recent immigrants and be less

prevalent among those who acquired some Canadian experience and upgraded their educational

credentials to Canadian standards after some years after arrival. However, after labour market

insertion has been successfully completed, migrants may face new challenges related to upward

social mobility, which may represent a new source of stress.

Although mediation by changes in maternal health behaviours and related risk factors is one

candidate explanation, other pathways are also possible. For example, chronic stressors, such as

low social position and discrimination, are more pervasive among minority women48 and may

represent alternative mechanisms leading to preterm birth among immigrants. Such hypothetical

pathways may involve health behaviours or not. Adoption of unhealthy health behaviours, such

as smoking and alcohol consumption, may be considered as a culturally acceptable way to cope

with chronic stress or adverse life events, imitated by immigrants. Even if migrants do not adopt

local unhealthy behaviours like these, exposure to stressful situations might precipitate preterm

labour via elevation of serum corticotrophin-releasing hormone (CRH), which creates a state of

hypervigilance or “arousal pathology”.48,49 Plasma CRH has emerged as a potential biological

marker for the prediction of preterm labour, although results obtained so far are not promising

enough to warrant its use as a routine clinical test.50 Different pathways may also intersect. It

129

has been suggested that genetic polymorphisms may interact with stress and obesity/overweight

to produce genitor-urinary infections, such as chorioamnionitis, that may trigger preterm

labour.51

The previous considerations, although speculative, draw attention to the potential complexity of

the observed association between time since migration and preterm birth. Whatever the pathways

are, future studies on this issue would greatly benefit from simultaneously obtaining data on

factors located at different steps in the hypothesized causal web (demographic, behavioural,

psychosocial, biological, and obstetrical). The acquisition and analysis of longitudinal data at the

individual level (e.g., before migration, at arrival, before and during pregnancy, and at delivery)

may allow the detection of changes in already established risk factors and their impact on

preterm birth. If obtaining detailed longitudinal data is not feasible, a first exploratory step would

be to compare recent and long-term immigrants, and try to explain their differences in preterm

birth by examining their differences in risk factors measured during pregnancy.

6.4.2. Is the association between time since migration and preterm birth merely a Canadian

phenomenon?

A second question is whether this direct association between length of stay after immigration and

risk of preterm birth is expected to be found in other settings. There is some evidence suggesting

that a positive association is likely to exist among migrants to Quebec25 and the U.S.,31,52

although it is not known whether such associations are characterized by a linear dose-response

pattern, such as ours. New comparisons, which will have to wait for similar studies to be

conducted elsewhere, may reveal similarities and differences between studies, which may help

identify underlying factors and potential pathways.

130

6.4.3. Is time since migration associated with other pregnancy-related outcomes?

The third question is whether the findings regarding duration of residence could be expected for

pregnancy-related outcomes other than preterm birth. For example, chapter 3 showed that

duration of residence affected gestational age but not fetal growth, as measured by SGA, which

is intriguing, since both outcomes are known to share common risk factors38,53 and

socioeconomic risk factors usually are associated with both outcomes in the same direction.54-56

One candidate risk factor to explain the association of duration of residence with preterm birth

(but not with “small for gestational age”) is maternal pre-pregnancy weight. Potential increases

in maternal pre-pregnancy body mass index with time spent in Canada (if confirmed) might be at

least partially responsible for counterbalancing the effect of other deleterious influences on small

for gestational age (e.g., smoking). In addition, the contribution of maternal pre-pregnancy

weight to gestational age is not clear in the literature 14,39,42 but could have influenced preterm

birth in our population. Although the testing of such hypotheses would need appropriate data that

are not easily available, the point here is to illustrate that further investigation of the potential

causal pathways behind these differential associations may advance our understanding of how

environmental exposures shape birth outcomes. A second candidate factor may be psychosocial

exposures, since some studies finding a positive association between different measures of stress

and preterm birth did not detect an association with intrauterine growth restriction.57-59

Another open question is whether time since migration is expected to predict infant mortality and

long-term consequences of preterm birth. It is reasonable to hypothesize that maternal duration

of residence since immigration would also be associated with those long-term childhood

outcomes that are known to follow after the occurrence of preterm birth, such as cerebral palsy,

although preterm birth may not be their cause but an endpoint resulting from common underlying

131

causes (e.g., pregnancy complications).5 That hypothesis could be tested using linkable

administrative health care databases, such as those used in this thesis, by following infants born

to immigrant women for some years after birth.

These kinds of questions would lead us to distinguishing which outcomes are affected similarly

by time since migration, and which outcomes are not, and why. Such inquiry could represent a

first step towards the elucidation of the influences operating on migrant women after arrival.

6.4.4. Why do some migrant groups experience poor outcomes and others do not?

The fourth question is why some migrant groups experience poor outcomes and others do not.

Unlike duration of residence that affected birth outcomes of immigrants from different world

regions almost uniformly in the same direction, differences between birth outcomes of

immigrants at a given duration of residence (e.g., recent immigrants by world region of origin in

chapter 4) are of great magnitude. The clarification of these differences would lead to the

investigation of the role of influences operating at multiple levels. At the country level, some

characteristics may explain differences between countries, such as the level of socioeconomic

and human development. Regional differences and social class dynamics within countries may

also be important in explaining why some people emigrate and why others do not and why, and

how such selection affects birth outcomes in the two groups. Finally, source and receiving

countries may differ in their patterns of emigration-immigration, thus shaping particular

immigrant experiences in many settings. The local environment of the receiving country may be

crucial in understanding immigrant health, particularly with increasing duration of stay.

Although the emphasis of this thesis was on urban neighbourhoods, other contexts such as the

physical environment, the workplace, and the family may be also relevant. Also, regional labour

and housing markets may attract or reject certain immigrants, especially at the city level. At the

132

individual level, maternal (and paternal) characteristics collected prior to migration may help

elucidate whether and to what extent so-called “healthy immigrant selection” is shaping the

outcomes of emigrants in their new home. Repeated measures of the same characteristics after

migration are desirable, although difficult to obtain, in order to track changes in risk factors and

their impact on the outcomes. In other words, longitudinal data should be ideally obtained on

time-dependent variables, at different points in time, in order to properly assess the impact of

migration and subsequent adaptation on immigrants’ health.

6.5. Concluding remark

Although this thesis can be labelled as a thesis on immigrants’ health, the relation between

migration and reproductive health goes beyond the health of minority groups. Indeed, the

offspring of Canada’s many immigrants are ‘Canadian-born’; and therefore, disparities in

reproductive outcomes among immigrants contribute directly to disparities in related childhood

and subsequent adult outcomes among the ‘non-immigrant’ population. They are therefore surely

worthy of increased Canadian research attention.

133

6.6. References

(1) Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology.

1999;10(1):37-48.




(4) Gagnon, A. J., Zimbeck, M., Zeitlin, J., and and the ROAM Collaboration. Migration to western

industrialized countries and perinatal health: A systematic review. Social Science & Medicine . 2008.

Ref Type: In Press

(5) Joseph KS. Theory of obstetrics: an epidemiologic framework for justifying medically indicated early

delivery. BMC Pregnancy Childbirth. 2007;7:4.


1241.

(7) Solis P, Pullum SG, Frisbie P. Demographic Models Of Birth Outcomes And Infant Mortality: An

Alternative Measurement Approach. Demography. 2000;37(4):489-498.



(9) Kierans WJ, Joseph KS, Luo ZC, Platt R, Wilkins R, Kramer MS. Does one size fit all? The case for

ethnic-specific standards of fetal growth. BMC Pregnancy Childbirth. 2008;8:1.

134

(10) Joseph KS, Demissie K, Kramer MS. Obstetric intervention, stillbirth, and preterm birth. Semin Perinatol.

2002;26(4):250-259.

(11) Joseph KS, Young DC, Dodds L et al. Changes in maternal characteristics and obstetric practice and recent

increases in primary cesarean delivery. Obstet Gynecol. 2003;102(4):791-800.

(12) Goldenberg RL, Rouse DJ. Prevention of premature birth. N Engl J Med. 1998;339(5):313-320.

(13) Saigal S, Doyle LW. An overview of mortality and sequelae of preterm birth from infancy to adulthood.

Lancet. 2008;371(9608):261-269.

(14) Committee on Understanding Premature Birth and Assuring Healthy Outcomes. Preterm Birth: Causes,

Consequences, and Prevention. First ed. Washington, D.C.: National Academies Press; 2007.

(15) Heaman MI, Sprague AE, Stewart PJ. Reducing the preterm birth rate: a population health strategy. J

Obstet Gynecol Neonatal Nurs. 2001;30(1):20-29.

(16) Petrou S, Mehta Z, Hockley C, Cook-Mozaffari P, Henderson J, Goldacre M. The impact of preterm birth

on hospital inpatient admissions and costs during the first 5 years of life. Pediatrics. 2003;112(6 Pt

1):1290-1297.

(17) Russell RB, Green NS, Steiner CA et al. Cost of hospitalization for preterm and low birth weight infants in

the United States. Pediatrics. 2007;120(1):e1-e9.

(18) Iams JD, Romero R, Culhane JF, Goldenberg RL. Primary, secondary, and tertiary interventions to reduce

the morbidity and mortality of preterm birth. Lancet. 2008;371(9607):164-175.

135

(19) Ananth CV, Vintzileos AM. Epidemiology of preterm birth and its clinical subtypes. J Matern Fetal

Neonatal Med. 2006;19(12):773-782.

(20) Hyman I, Dussault G. Negative consequences of acculturation on health behaviour, social support and

stress among pregnant Southeast Asian immigrant women in Montreal: an exploratory study. Can J Public

Health. 2000;91(5):357-360.

(21) Hyman I. Setting the stage: reviewing current knowledge on the health of Canadian immigrants: what is the

evidence and where are the gaps? Can J Public Health. 2004;95(3):I4-I8.

(22) Pollitt RA, Kaufman JS, Rose KM, ez-Roux AV, Zeng D, Heiss G. Cumulative life course and adult

socioeconomic status and markers of inflammation in adulthood. J Epidemiol Community Health.

2008;62(6):484-491.

(23) Cummins S, Curtis S, ez-Roux AV, Macintyre S. Understanding and representing 'place' in health research:

a relational approach. Soc Sci Med. 2007;65(9):1825-1838.

(24) Stansfeld S, Candy B. Psychosocial work environment and mental health--a meta-analytic review. Scand J

Work Environ Health. 2006;32(6):443-462.

(25) Hyman I, Dussault G. The effect of acculturation on low birthweight in immigrant women. Can J Public

Health. 1996;87(3):158-162.


Public Health. 1989;79(9):1263-1267.

(27) Hawkins SS, Lamb K, Cole TJ, Law C. Influence of moving to the UK on maternal health behaviours:

prospective cohort study. BMJ. 2008;336(7652):1052-1055.

136

(28) Detjen MG, Nieto FJ, Trentham-Dietz A, Fleming M, Chasan-Taber L. Acculturation and cigarette

smoking among pregnant Hispanic women residing in the United States. Am J Public Health.

2007;97(11):2040-2047.

(29) Cobas JA, Balcazar H, Benin MB, Keith VM, Chong Y. Acculturation and low-birthweight infants among

Latino women: a reanalysis of HHANES data with structural equation models. Am J Public Health.

1996;86(3):394-396.

(30) Goel MS, McCarthy EP, Phillips RS, Wee CC. Obesity among US immigrant subgroups by duration of

residence. JAMA. 2004;292(23):2860-2867.



(32) McDonald JT, Kennedy S. Is migration to Canada associated with unhealthy weight gain? Overweight and

obesity among Canada's immigrants. Soc Sci Med. 2005;61(12):2469-2481.

(33) Ray JG, Vermeulen MJ, Schull MJ, Singh G, Shah R, Redelmeier DA. Results of the Recent Immigrant

Pregnancy and Perinatal Long-term Evaluation Study (RIPPLES). CMAJ. 2007;176(10):1419-1426.

(34) Henriksen TB, Wilcox AJ, Hedegaard M, Secher NJ. Bias in studies of preterm and postterm delivery due

to ultrasound assessment of gestational age. Epidemiology. 1995;6(5):533-537.

(35) Jain T, Hornstein MD. Disparities in access to infertility services in a state with mandated insurance

coverage. Fertil Steril. 2005;84(1):221-223.

137

(36) Schieve LA, Ferre C, Peterson HB, Macaluso M, Reynolds MA, Wright VC. Perinatal outcome among

singleton infants conceived through assisted reproductive technology in the United States. Obstet Gynecol.

2004;103(6):1144-1153.

(37) Goldenberg RL, Culhane JF, Iams JD, Romero R. Epidemiology and causes of preterm birth. Lancet.

2008;371(9606):75-84.

(38) Kramer MS. Determinants of low birth weight: methodological assessment and meta-analysis. Bull World

Health Organ. 1987;65(5):663-737.

(39) Cnattingius S, Bergstrom R, Lipworth L, Kramer MS. Prepregnancy weight and the risk of adverse

pregnancy outcomes. N Engl J Med. 1998;338(3):147-152.

(40) Haeri S, Guichard I, Baker AM, Saddlemire S, Boggess KA. The effect of teenage maternal obesity on

perinatal outcomes. Obstet Gynecol. 2009;113(2 Pt 1):300-304.

(41) Hendler I, Goldenberg RL, Mercer BM et al. The Preterm Prediction Study: association between maternal

body mass index and spontaneous and indicated preterm birth. Am J Obstet Gynecol. 2005;192(3):882-886.

(42) Nohr EA, Bech BH, Vaeth M, Rasmussen KM, Henriksen TB, Olsen J. Obesity, gestational weight gain

and preterm birth: a study within the Danish National Birth Cohort. Paediatr Perinat Epidemiol.

2007;21(1):5-14.

(43) Smith GC, Shah I, Pell JP, Crossley JA, Dobbie R. Maternal obesity in early pregnancy and risk of

spontaneous and elective preterm deliveries: a retrospective cohort study. Am J Public Health.

2007;97(1):157-162.

138

(44) Siega-Riz AM, Siega-Riz AM, Laraia B. The implications of maternal overweight and obesity on the

course of pregnancy and birth outcomes. Matern Child Health J. 2006;10(5 Suppl):S153-S156.

(45) Torloni MR, Betran AP, Horta BL et al. Prepregnancy BMI and the risk of gestational diabetes: a

systematic review of the literature with meta-analysis. Obes Rev. 2008.

(46) Chu SY, Kim SY, Schmid CH et al. Maternal obesity and risk of cesarean delivery: a meta-analysis. Obes

Rev. 2007;8(5):385-394.

(47) Meis PJ, Goldenberg RL, Mercer BM et al. The preterm prediction study: risk factors for indicated preterm

births. Maternal-Fetal Medicine Units Network of the National Institute of Child Health and Human

Development. Am J Obstet Gynecol. 1998;178(3):562-567.

(48) Rich-Edwards JW, Grizzard TA. Psychosocial stress and neuroendocrine mechanisms in preterm delivery.

Am J Obstet Gynecol. 2005;192(5 Suppl):S30-S35.

(49) McCubbin JA, Lawson EJ, Cox S, Sherman JJ, Norton JA, Read JA. Prenatal maternal blood pressure

response to stress predicts birth weight and gestational age: a preliminary study. Am J Obstet Gynecol.

1996;175(3 Pt 1):706-712.

(50) McLean M, Smith R. Corticotrophin-releasing hormone and human parturition. Reproduction.

2001;121(4):493-501.

(51) Crider KS, Whitehead N, Buus RM. Genetic variation associated with preterm birth: a HuGE review. Genet

Med. 2005;7(9):593-604.

139

(52) Crump C, Lipsky S, Mueller BA, Crump C, Lipsky S, Mueller BA. Adverse birth outcomes among

Mexican-Americans: are US-born women at greater risk than Mexico-born women? Ethnicity & Health.

1999;4(1-2):29-34.

(53) Lang JM, Lieberman E, Cohen A. A comparison of risk factors for preterm labor and term small-for-

gestational-age birth. Epidemiology. 1996;7(4):369-376.

(54) Luo ZC, Kierans WJ, Wilkins R, Liston RM, Mohamed J, Kramer MS. Disparities in birth outcomes by

neighborhood income: temporal trends in rural and urban areas, british columbia. Epidemiology.

2004;15(6):679-686.

(55) Luo ZC, Wilkins R, Kramer MS. Effect of neighbourhood income and maternal education on birth

outcomes: a population-based study. CMAJ. 2006;174(10):1415-1420.

(56) Joseph KS, Liston RM, Dodds L, Dahlgren L, Allen AC. Socioeconomic status and perinatal outcomes in a

setting with universal access to essential health care services. CMAJ. 2007;177(6):583-590.

(57) Copper RL, Goldenberg RL, Das A et al. The preterm prediction study: maternal stress is associated with

spontaneous preterm birth at less than thirty-five weeks' gestation. National Institute of Child Health and

Human Development Maternal-Fetal Medicine Units Network. Am J Obstet Gynecol. 1996;175(5):1286-

1292.

(58) Rondo PH, Ferreira RF, Nogueira F, Ribeiro MC, Lobert H, Artes R. Maternal psychological stress and

distress as predictors of low birth weight, prematurity and intrauterine growth retardation. Eur J Clin Nutr.

2003;57(2):266-272.

(59) Nordentoft M, Lou HC, Hansen D et al. Intrauterine growth retardation and premature delivery: the

influence of maternal smoking and psychosocial factors. Am J Public Health. 1996;86(3):347-354.

140

Appendices

Appendix 2.A. Search strategy

Original search strategy carried out by ROAM1

Criteria for considering studies for review

Study outcomes: Reports were excluded if the outcome under study was not directly related to

perinatal health indicators commonly cited by national or international bodies (e.g., EURO-

PERISTAT),2 or to outcome differences specific to pregnant migrants such as infectious disease

and smoking/ drug/ alcohol use.

Type of exposure: Migration was the exposure of interest. Therefore, any report of women who

had migrated was initially included. Reports were subsequently excluded if international cross-

border movement was unlikely (thus ‘protectorates’ such as Puerto Rico did not meet our

definition of migrant, nor did second-generation populations.)

Type of study designs: We excluded case studies, clinical reports, reports without a comparison

group, and reports in which the results of the migrant group(s) were not presented separately

from the comparison group.

Study population: Migrant women in ‘western industrialized’ countries were included. Those in

refugee camps were excluded.

An author search of ROAM collaborators was also conducted. Searches were supplemented with

bibliographic citation hand-searches of included literature published from 2004 onward. Other

relevant literature referred to the authors was also reviewed. Government documents or ‘grey

literature’ were not reviewed and no attempts were made to contact authors of published works

due to the volume of literature under consideration.

141

No language exclusions were routinely applied. French, Spanish, Swedish, and Italian- language

articles were reviewed by members of ROAM or their associates. Review of other non-English

language articles was restricted to their English abstracts. On two occasions, an English abstract

was not available and the article was not reviewed.

Search strategy: Electronic literature databases from 1995 through September 2006 were

searched using Ovid (version 10.5.1) in the following order: Medline, Health Star, Embase, and

PsychInfo. The search strategy was developed in conjunction with a McGill University Health

Sciences librarian and was as follows:

1 exp emigration/ and immigration.mp. 2 *Ethnic Groups/ 3 *Minority Groups/ 4 (migration history or premigration).tw. 5 exp refugees/ or refugee$.tw. 6 migrant.tw. 7 perinatal$.tw. 8 exp pregnancy/ or pregnancy outcomes.mp. or pregnancy complications.mp. or pregnancy.tw. 9 equity.tw. 10 exp delivery of healthcare/ 11 exp maternal health service$/ 12 exp social support/ or social isolation.tw. 13 exp prejudice/ or (racism or prejudice).tw. 14 1 or 2 or 3 or 4 or 5 or 6 15 14 and 8 16 9 or10 or 11 or 12 or 13 17 14 and 16 18 7 and 17 19 15 or 17 20 limit 19 to yr="1995 - 2006"

Updated search: The above search identified 78 studies relevant for the review of low

birthweight and preterm birth by migrant subgroups.

142

The same search was repeated in early December 2007 to include articles published up until

November 2007. One article written in Serbian was translated into English. Back references

were also checked and potential articles pointed out by ROAM members were also considered.

Thus, six new papers were considered but only four met inclusion criteria, totalling 82.

References

(1) Gagnon, A. J., Zimbeck, M., Zeitlin, J., and the ROAM Collaboration. Migration to western industrialized

countries and perinatal health: A systematic review. Social Science & Medicine (In press), 2009.

(2) Zeitlin J, Wildman K, Breart G et al. PERISTAT: indicators for monitoring and evaluating perinatal health in

Europe. European Journal of Public Health. 2003;13(3 Suppl):29-37.

143

Appendix 3.A. Data sources

Discharge Abstracts Database (DAD-CIHI): The DAD contains information on the individual

health number, address information, birth date, admission and discharge dates, 16 diagnostic

codes, 10 procedure codes from fiscal years 1988/1989 to 2001/2002 (ICD-9) and 25 diagnostic

and 20 procedure codes from fiscal year 2002/2003 onwards (ICD-10-CA), among many other

variables describing the nature of the hospital services. CIHI data are available at ICES thanks to

an agreement held between the Institute for Clinical Evaluative Sciences (ICES) and the Ontario

Ministry of Health and Long-Term Care (MOHLTC). The DAD has three main limitations. First,

out-of-hospital births are not captured, which would represent about 1.1% of all births.1 Second,

information on gestational age and recent parity were not captured until fiscal year 2002/2003,

when a data-system redevelopment took place. Since fiscal 2002/2003, gestational age is

recorded in completed weeks and is derived from medical charts, thus representing the best

clinical estimate of gestation, which includes both ultrasound- and LMP-based estimates.1 Third,

the mother and child records were unrelated until fiscal year 2002/2003, when a link was created

through the inclusion of both the mother and child health number in each record. This study

therefore had to rely on a probabilistic linkage, which is known to capture close to 95% of births

up until fiscal 2001/2002. Despite the deterministic link between the mother and child records

was available since 2002/2003, the matching rate (under 80% in 2002/2003 and under 90% in

2003/2004) was not as good as the one obtained with the probabilistic approach. Therefore, the

probabilistic was used throughout the study period.

The approach of extracting data on both live births and mothers from the DAD was evaluated by

members of the Canadian perinatal surveillance system group,2 proving to have excellent

coverage and accurate rates for Ontario. More sensitive methods have been developed in order to

144

capture some births missed by the Wen’s approach.3,4 An ICES re-abstraction study based on the

DAD (2002/03 and 2003/04 fiscal years) found excellent percent agreement on variables key for

this study (99% in sociodemographic variables, 100% in liveborn infants according to place of

birth and 92.3% in comorbid diagnoses such as preterm delivery).5 The meaning of each variable

was interpreted following the CIHI Abstracting Manual (1995)6 and DAD Data Quality

Documentation.7,8 Diagnostic and procedure codes correspond to the International Classification

of Diseases - 9th Revision (ICD-9)9 and to the Canadian Classification of Diagnostic,

Therapeutic, and Surgical Procedures (CCP) up until fiscal 2002/2003, when the enhanced

Canadian version of the 10th revision (ICD-10-CA) and the Canadian Classification of Health

Interventions (CCI) were adopted in Ontario.8,10

Landed Immigrant Data System (LIDS): The LIDS dataset spans from calendar year 1985 to

2000. The Ontario LIDS is composed of the landing records of all legal immigrants whose

intended destination was Ontario. A probabilistic linkage with the provincial administrative

health care database (RPDB) was carried out at ICES using a special algorithm based on

surname, given names, sex, and date of birth, and complemented by manual review of non-

matched records.11 The information contained in each record is based on the documentation

provided by the migrants during their application process: age, gender, marital status, country of

birth, citizenship, and last permanent residence, mother tongue, intended destination; immigrant

category, special program codes, principal applicant code, employment status; intended

occupation, years of schooling, level of education, knowledge (self–assessed) of an official

language. All of the information taken from the Landing Record was recorded as of the date of

issue of the landing visa and is retained intact regardless of the year of observation.

145

The Registered Persons Data Base (RPDB): Upon approval for Ontario health coverage (Ontario

Health Insurance Plan - OHIP), client registration and identification information is entered onto

the Registered Persons Data Base (RPDB), maintained by the Ministry of Health since 1990. The

RPDB contains information on individuals such as their surname, given name, health number,

gender, date of birth, place of residence, and the dates of the start and end of eligibility for the

provincial health coverage for all people residing in Ontario. This database was used to extract

information on new registrants to OHIP, as a proxy for recent immigration (i.e., within the last

five years).

Details of the linkage between the LIDS and the RPDB

Because the probabilistic linkage of the LIDS with the provincial administrative health care

database (RPDB) was done provincially (84.4% match), the linkage does not capture those

international migrants residing in Ontario who came into Canada via any port of entry outside

Ontario and moved in afterwards. Conversely, migrants who arrived to Ontario and moved out of

the province shortly after arrival may not be captured by the linked dataset. A initial validation

study 12 concluded that there was little bias for linkage based on the percent linked for different

predictor variables included in the LIDS. There were differences in the following categories: 1)

Landing year: The percent linked diminished to a low of 72% prior to 1990, presumably due to

the creation of new electronic health cards in 1991; 2) Visa category: Business class had a

relatively low linkage (72.3%), most likely because immigrants in this category were more likely

to continue to live in their home country, such as those from Hong Kong who transferred power

to China after becoming landed immigrants; 3) Education: Slightly lower linkage for Masters

and PhD degrees; 4) Country: There were no great variations across countries represented by

146

more than 100 migrants. Hong Kong had one of the lowest linkage rates (66.5%), probably due

to their Visa Category.

The lower linkage rates for migrants in the business class category, holding post-secondary

education, and from Hong Kong suggest that non-linked migrants were more resourceful

individuals that may have abandoned Ontario shortly after arrival pursuing better prospects

somewhere else. If this is true, the low linkage rate among these groups would not result in

misclassification because they are not supposed to be counted as non-immigrant residents of

Ontario. However, potential bias may arise if many immigrants came to Ontario after landing in

other provinces, since this type of immigrants could not be captured by the linkage. Such bias

would affect comparisons between immigrants and non-immigrants towards the null effect, since

immigrants whose intended destination was not Ontario would be ‘false non-immigrants’.

However, there is no bias when comparisons are made between immigrant groups, because the

reference group is an internal group of immigrants.

Canadian Censuses 1996 and 2001: Contextual data at the census tract (CT) level were merged

with the above datasets through the Postal Code Conversion File Plus (PCCF+) from Statistics

Canada. The PCCF+ is a SAS program that assigns each 6-digit postal code to enumeration

areas/dissemination areas, census tracts and other levels of census geography, and is routinely

updated to reflect changes in census geography.

147

Table 3.A.1. Variable definitions

Variable Definition Scale Data source Availability

Outcomes

Birthweight (BW) In grams (>=500 and <=6000) Continuous DAD-CIHI 1988/89 – 2006/07

Low birthweight

(LBW)

Infant weighting <2500 grams Binary DAD-CIHI 1988/89 – 2006/07

Very low

birthweight

(VLBW)

Infant weighting < 1,500 grams Binary DAD-CIHI 1988/89 – 2006/07

Moderately low

birthweight

(MLBW)

Infant weighting 1,500-2,499 grams Binary DAD-CIHI 1988/89 – 2006/07

Preterm birth

(PTB)

Infant born before the 37th completed week

of gestation (at maternal hospital

admission)

Binary DAD-CIHI 1988/89 – 2001/02

based on ICD-9

codes

2002/03 – 2006/07

based on weeks of

gestation

Very preterm birth

(VPTB)

Infant born before 32 weeks Binary DAD-CIHI 2001/02 – 2006/07

Moderately

preterm birth

(MPTB)

Infant born between 32 weeks or more but

before 37 weeks

Binary DAD-CIHI 2001/02 – 2006/07

Small preterm

birth (SPTB)

Infants having both LBW and PTB Binary DAD-CIHI 1988/89 – 2006/07

Small for

gestational age

(SGA)13

Gender and gestational-age specific

birthweight below the 10th percentile of the

most recent published sex-specific

Canadian reference values based on

infants born in 1994–96

Binary DAD-CIHI 2001/02 – 2006/07

Independent

variables

Immigrant status Presence of a landing record Binary LIDS 1985 - 2000

Immigrant status

by length of stay

Presence of a landing record 0-4 years to

delivery (recent immigrant), 5-9 years, 10-

14 years, more than 14 years (long-term

resident)

Categorical LIDS and

DAD-CIHI

1985 – 2006/07

Length of stay Time from arrival to delivery (in days).

Based on a combination of arrival and

landing date.

Continuous LIDS and

DAD-CIHI

1985 - 2006/07

World regions Groupings by country of birth based on Categorical LIDS 1985 - 2000

148

Variable Definition Scale Data source Availability

international classifications (World Bank,

UNICEF, United Nations)

Area-based

material

deprivation

Neighbourhood indexes obtained by

principal component analysis (Material

deprivation, residential instability,

dependency, ethnic diversity)

Continuous CENSUS 1996, 2001

Covariates

Infant sex Female, male Categorical DAD-CIHI 1988/89 – 2006/07

Maternal age Maternal age at delivery (15-20, 20-24, 25-

29[ref], 30-34, 35-39, 40-55)

Categorical DAD-CIHI 1988/89 – 2006/07

Parity Number of previous live births Discrete DAD-CIHI 2002/03 – 2006/07

Previous preterm

delivery

Yes, no [ref] Binary DAD-CIHI 2002/03 – 2006/07

Marital status at

arrival

Married/common law [ref],

Single/divorced/separated

Categorical LIDS 1985 - 2000

Immigrant class Economic class [ref], Family class,

Refugees


Knowledge of

English/French Yes [ref], no Binary LIDS 1985 - 2000

Maternal

education at

arrival

0 to 9, 10 to 12 years of schooling, any

post-secondary diploma [ref]


High School

graduation

Yes [ref], no Binary LIDS 1985 - 2000

Pre-existing

maternal illnesses

and pregnancy

complications

Yes, no [ref] Binary DAD-CIHI 1988/89 – 2006/07

Recent

registration to

OHIP

Start of eligibility for OHIP coverage within

a five-year period prior to delivery after April

1, 1991

Binary RPDB 1995/96 – 2006/07

References

(1) Canadian Institute for Health Information. Too Early, Too Small: A Profile of Small Babies Across

Canada. 2009. Ottawa, Ont, CIHI.

149

(2) Wen SW, Liu S, Marcoux S, Fowler D. Uses and limitations of routine hospital admission/separation

records for perinatal surveillance. Chronic Dis Can. 1997;18(3):113-119.

(3) Health Canada. Canadian Perinatal Health Report, 2003. Cat. No. H49-142/2003E. 2003. Ottawa, Minister

of Public Works and Government Services Canada.



(5) Juurlink D, Preyra, C, Croxford, R, Chong, A, Austin, P, Tu, J, and Laupacis, A. Canadian Institute for

Health Information Discharge Abstract Database: a validation study. 2006. Toronto, Institute for Clinical

Evaluative Sciences (ICES).

(6) Canadian Institute for Health Information (CIHI). Abstracting Manual. 1995. Canadian Institute for

Health Information (CIHI).





(9) World Health Organization. International Classification of Diseases, Injuries and Causes of Death. 9th

revision. Geneva: The Organization; 1979.

(10) Canadian Institute for Health Information. Final Report. The Canadian Enhancement of ICD-10

(International Statistical Classification of Diseases and Related Health Problems, Tenth Revision). 2001.

Ottawa, Ont., CIHI.

150

(11) Desmeules, M., McDermott, S., Cao, Z., Manuel, D., Kazanjian, A., Vissandee, B., Ruddick, E., Kleiwer,

E., Mao, Y., and Gold, J. Immigrant Health and Health Care Utilization in Canada: Phase II of the National

Immigrant Health Initiative. 2004.

(12) Cernat, G, Wall, C, Iron, K, and Manuel, D. Initial Validation of Landed Immigrant Data System (LIDS)

with the Registered Person's Database (RPDB) at ICES. 2002.



151

Appendix 3.B. Measurement of Stillbirths and Multiple Births Using the Discharge

Abstract Database

Approaches to extracting stillbirths and multiple births from the DAD vary according to the unit

of analysis, and the use of diagnostic codes. For example, one ICES report 1 used ICD-9/10

codes in the mothers’ records presumably because the unit of analysis of the report was the

mother and not the child. In the same vein, two recent CIHI reports used information contained

in the child records because the report focused on the infant.2,3

Stillbirths

One CIHI report used ICD codes to identify multiple births but not to identify stillbirths.3

Instead, the report used a variable “hospital entry code = S (stillborn)”. However, this

information has not been consistently sent to ICES by CIHI (Alex Kopp, personal

communication). Therefore, we relied solely on the diagnostic codes. Table 1 and Figure 1 show

the stillbirth ‘rates’ from 1998 to 2006 using different approaches: a) information in the maternal

records only; b) information in the infant records only; c) information in either the maternal or

infant records; and d) the Vital Statistics estimates, added as an external comparison.4,5 The same

definition of fetal death rate (crude) (i.e., number of stillbirths per 1,000 total births (live births

and stillbirths)) was used for all estimates.

152

Table 3.B.1. Measurement of stillbirths (fetal deaths per 1,000 total births) in the Ontario DAD-CIHI at

ICES (fiscal 1988/1989-2006/2007), by method, and in the Ontario Vital Statistics (calendar 1991-2004)

(a) (b) (c) (d)

Year Mothers’

ICD

Infants'

ICD

Either

ICD

Vital

Stats

1988 5.7 2.0 6.4

1989 6.4 1.6 7.0

1990 5.0 1.6 5.7

1991 5.2 2.6 6.9 5.3

1992 5.1 1.9 6.1 7.1

1993 4.7 2.8 6.1 6.5

1994 4.3 1.6 4.8 6.4

1995 4.3 1.7 5.2 6.6

1996 4.9 3.2 6.6 6.4

1997 4.8 2.9 5.7 6.6

1998 5.4 2.1 6.2 6.4

1999 5.7 2.7 6.8 6.3

2000 5.2 2.6 6.5 6.4

2001 5.6 3.6 7.7 6.3

2002 10.0 0 10.0 6.3

2003 6.8 0 6.8 7.3

2004 7.0 0 7.0 6.3

2005 5.2 0 5.2

2006 4.5 0 4.5

153

Figure 3.B.1. Measurement of stillbirths in the Ontario DAD-CIHI at ICES (fiscal 1988/1989-2006/2007),

by method, and in the Ontario Vital Statistics (calendar 1991-2004)

The use of infant records only clearly underestimates stillbirth counts and gives no information

after the implementation of the ICD-10, supposedly because of a change in the coding (see Table

3.B.4 below). The use of maternal records provides a better estimate that is still below the Vital

statistics’ estimates, with the exception of the first years following the implementation of the

ICD-10 where the sudden jump in the rates is suggestive of an artefact. Such interruption of the

trends from the ICD-9 to the ICD-10 is common to many other outcomes (Alex Kopp, personal

communication). The combination of both the maternal and infant records increases the rate for

most years, indicating that the codes in the infant records capture some stillborns not included in

the maternal records. This combination approaches Vital Stats data reasonably well for some

years but it is generally below. However, in theory it is expected that the measurement of

stillbirths would result in higher rates than those based on the Vital Stats. First, the exclusion of

154

home births from the CIHI should have a modest impact on these rates as stillbirths and multiple

births typically occur in hospital (about half of all stillbirths result from multiple pregnancies)

and the healthy singletons born outside hospital are not included in the denominator. Second, the

CIHI data are likely to capture extremely small babies which may not get registered in the Vital

Statistics data. For instance, the coders at the hospital code the CIHI records but the physicians

are those who likely determine whether a stillbirth certificate is to be filled out or not (K.S.

Joseph, personal communication).

Therefore, based on these considerations, the best possible estimate of stillbirths from the DAD

to be used in this thesis is to combine maternal and infant records.

Multiple births

Table 3.B.2 and Figure 3.B.2 show the multiple birth percent from 1998 to 2006 using

different approaches: a) information in the maternal records only; b) information in the infant

records only; c) information in either the maternal or infant records; d) counting the number of

child records per delivery episode; e) a combination of (c) and (d); and f) the Vital Statistics

estimate, added as an external comparison. The same definition was used for all estimates (i.e.,

number of multiple births per 100 births, including stillbirths and live births).5

155

Table 3.B.2. Measurement of multiple births (multiple births per 100 total births) in the Ontario DAD-CIHI

at ICES (fiscal 1988/1989-2006/2007), by method, and in the Ontario Vital Statistics (calendar 1991-

2004)

(a) (b) (c) (d) (e) (f)

Maternal

ICD

Infants'

ICD

Either

ICD

Count

method

Either count

or either ICD

Vital

Stats

1988 1.47 1.08 1.53 3.13 3.96

1989 1.53 1.11 1.58 2.54 3.33

1990 1.63 1.27 1.70 2.16 2.82

1991 2.20 1.71 2.27 2.74 2.96 2.0

1992 2.41 1.86 2.46 2.73 2.84 2.1

1993 2.42 2.15 2.49 2.69 2.75 2.1

1994 2.51 2.27 2.58 2.76 2.82 2.4

1995 2.40 2.17 2.46 2.69 2.75 2.4

1996 2.64 2.43 2.69 2.86 2.95 2.6

1997 2.77 2.53 2.80 2.90 2.98 2.7

1998 2.91 2.71 2.96 3.08 3.16 2.8

1999 2.92 2.68 2.95 3.06 3.16 2.9

2000 2.90 2.71 3.05 3.14 3.22 2.9

2001 3.07 2.81 3.13 3.25 3.35 3.0

2002 3.17 2.94 3.24 3.38 3.45 3.2

2003 3.24 3.02 3.27 3.43 3.48 3.2

2004 3.30 3.10 3.36 3.44 3.53 3.2

2005 3.27 3.07 3.31 3.40 3.50

2006 3.36 3.13 3.41 3.46 3.49

156

Figure 3.B.2. Measurement of multiple births in the Ontario DAD-CIHI at ICES (fiscal 1988/1989-

2006/2007), by method, and in the Ontario Vital Statistics (calendar 1991-2004)

Again, in theory it is expected that the estimates from the DAD will be higher than those based

on the Vital Stats because: i) multiple births are much more likely to occur in hospitals and home

births are represented by mostly uncomplicated single deliveries that are not included in the

denominator, ii) The main form for the registration of a live birth is completed by the parents,

who are responsible for filing it with the local registrar and failure to do this by the parents is one

cause of underegistration of births in the Ontario Vital Stats system and it might be higher when

multiple births are involved, iii) Over-coverage in the Vital Stats is minimal. Births to non-

resident women in Canada are registered but are excluded from most tabulations. Duplicate birth

registrations are identified as part of the regular processing operations on each provincial and

territorial subset, as well as additional inter-provincial checks, and comparisons between the

birth and stillbirth databases for multiple births. Possible duplicate registrations are checked

157

against microfilmed registrations or optical images, or by consulting with the provinces and

territories. In contrast, the probabilistic linkage of maternal and infant records may produce some

duplicates in a few cases where the criteria (i.e., postal code, maternal admission date, hospital)

are met by more than one delivery. Although the program developed at ICES includes a cleaning

algorithm, some duplicates remained. This is particularly evident when we consider the count

method (i.e., counting the number of infant records per maternal delivery episode), which is

consistently higher than the combination of maternal and infant ICD codes. A quick look at

Table 3.B.3 suggests that the count method is not a good approach to measure multiple births

since it overcounts some singletons matched to more than one mother, thus producing duplicate

records, as a result of the probabilistic linkage.

Table 3.B.3. Percentiles of birthweight (in grams) by method

5th

percentile

25th

percentile

50th

percentile

75th

percentile

95th

percentile

Multiple births by

ICD 1105 2070 2495 2850 3350

Singleton births by

ICD 2540 3105 3435 3770 4286

Multiple births by

the count method

but not by ICD

2360 3045 3400 3760 4290

If the count method identified only true multiples we would expect that the birthweight

distribution of those identified as multiple by the count method but not by the ICD codes’

method would be approximately similar to that of the rest of the multiples identified by the ICD

codes’ method. However, as seen in Table 3.B.3 this is not the case. These additional infants are

most likely false multiple births because their birthweight distribution is quite similar to that of

158

the singleton births. In conclusion, these exploratory analyses suggest that the best estimate of

multiple births obtained from the DAD, as available for this thesis, is the one based on a

combination of infant and maternal ICD codes.

Table 3.B.4. List of ICD codes for stillbirths according to Revision and unit of analysis

Stillbirths ICD-9 Description ICD-10 Description

Infants’

records

V320 Twin, mate stillborn, born in

hospital

P95 Fetal death of unspecified cause

V321 Twin, mate stillborn, born before

admission

V322 Twin, mate stillborn, born outside

hospital

V350 Other multiple, mates all stillborn,

born in hospital


born before admission


born outside hospital

Mothers’

records

6564 Intrauterine death O364 Maternal care for intrauterine

death

V271 Single stillborn Z371 Single stillbirth

V274 Twins, both stillborn Z374 Twins, both stillborn

V277 Other multiple birth, all stillborn Z377 Other multiple births, all stillborn

159

Table 3.B.5. List of ICD codes for multiple births according to Revision and unit of analysis

Multiple

births

ICD-9 Description ICD-10-

CA

Description

Infants’

records

7594 Conjoined twins Q894 Conjoined twins

V310 Twin, mate liveborn, born in

hospital

V311 Twin, mate liveborn, born before

admission

V312 Twin, mate liveborn, born outside

hospital

V330 Twin, unspecified, born in hospital Z383 Twin, born in hospital

V331 Twin, unspecified, born before

admission

Z384 Twin, born outside hospital

V332 Twin, unspecified, born outside

hospital

Z385 Twin, unspecified as to place of

birth

V340 Other multiple, mates all liveborn,

born in hospital


born before admission


born outside hospital

V360 Other multiple, mates live- and

stillborn, born in hospital


stillborn, born before admission


stillborn, born outside hospital

V370 Other multiple, unspecified, born

in hospital

Z386 Other multiple, born in hospital


before admission

Z387 Other multiple, born outside

hospital


outside hospital

Z388 Other multiple, unspecified as to

place of birth

Mothers’

records

6510 Twin pregnancy O300 Twin pregnancy

6511 Triplet pregnancy O301 Triplet pregnancy

6512 Quadruplet pregnancy O302 Quadruplet pregnancy

6513 Twin pregnancy with fetal loss

and retention of one fetus

160

Multiple

births

ICD-9 Description ICD-10-

CA

Description

6514 Triplet pregnancy with fetal loss

and retention of one or more

fetus(es)

6515 Quadruplet pregnancy with fetal

loss and retention of one or more

fetus(es)

6516 Other multiple pregnancy with

fetal loss and retention of one or

more fetus(es)

6517 Multiple gestation following

(elective) fetal reduction

6518 Other specified multiple gestation O308 Other multiple gestation

6519 Unspecified multiple gestation O309 Multiple gestation, unspecified

6526 Multiple gestation with

malpresentation of one fetus or

more

O325 Maternal care for multiple

gestation with malpresentation of

one fetus or more

6605 Locked twins O661 Obstructed labour due to locked

twins

7615 Multiple pregnancy O848 Other multiple delivery

O849 Multiple delivery, unspecified

V272 Twins, both liveborn Z372 Twins, both liveborn

V275 Other multiple birth, all liveborn Z375 Other multiple births, all liveborn

O840 Multiple delivery, all spontaneous

O841 Multiple delivery, all by forceps

and vacuum extractor

O842 Multiple delivery, all by

caesarean section

References

(1) Maaten S, Guttman A, Kopp A, Janda M, Jaakkimainen L. Care of women during pregnancy and

childbirth. Primary Care in Ontario. Toronto, Ontario: Institute for Clinical Evaluative Sciences, 2006: 15-

34.

(2) Canadian Institute for Health Information. Too Early, Too Small: A Profile of Small Babies Across Canada.

2009. Ottawa, Ont, CIHI.

161

(3) Canadian Institute for Health Information. Analysis in Brief: Factors Associated With Low Birth Weight,

2002-2003 to 2006-2007. [Canadian Institute for Health Information]. 2009. Available at:

http://secure.cihi.ca/cihiweb/dispPage.jsp?cw_page=RC_2309_E&cw_topic=2309.

(4) Health Canada. Canadian Perinatal Health Report, 2003. Cat. No. H49-142/2003E. 2003. Ottawa, Minister of

Public Works and Government Services Canada.

(5) Public Health Agency of Canada. Canadian Perinatal Health Report, 2008 Edition. Public Health Agency of

Canada . 2008. Ottawa, Public Health Agency of Canada.

162

Appendix 3.C. Flowchart Data Exclusions

Flow chart: Selection criteria for the study population

N Selection criteria

2,361,427 Birth records matched to a mother fiscal 1988-2006

↓ Keep only those babies born after ‘31MAR2002’ (New CIHI data)

630,925

↓ Keep those infant born within 2001 CMAs (1,913 records had

missing CMAs and were deleted)

491,324

↓ Exclude multiple births and stillbirths (N=16,710)

474,614

↓ Exclude infants born to mothers that had never been registered to

OHIP up until ‘31MAR2001’(last date in the LIDS=‘31DEC2000’ +

3-month waiting period). These may be migrants arriving after

‘31DEC2000’ (last date in the LIDS) or interprovincial migrants. As

these do not appear in the available LIDS data, we do not want them

to be misclassified as ‘non-immigrants’ (N=74,961)

399,653

↓ Exclude live births weighting less than 500 (due to increased

registration of near viable births) and more than 6000 g (and likely

data errors) (N=360)

399,293

↓ Exclude records with missing data on the outcomes or in gestational

age (N=125)

399,168

↓ Exclude records with gestational age lower than 22 weeks and higher

than 43 weeks (N=72)

399,096

↓ Exclude records with missing data on infant sex, maternal age or

163


parity (N=54)

399,042

↓ Exclude records with missing data on immigration characteristics

(N=576)

398,466

↓ Exclude immigrants classified as “OTHER” (i.e., other than ‘family’,

‘economic, or ‘refugee’) (N=487)

397,979

↓ Exclude records to which census information could not be assigned

(N=509)

397,470 Population size for analyses (immigrants = 83,233 and Canadian-

born/long-term residents = 314,237)

164

Appendix 3.D. Covariate Adjustment Based on Directed Acyclic Graphs (DAGs)

Here we present the rationale behind the covariate selection for the models shown in Table 3.3,

Chapter 3. The conventional criteria for confounding control works well when the effect of

exposure on disease is influenced by a third variable. However, these criteria may be inadequate

when multiple confounders are considered simultaneously. The theory of directed acyclic graphs

(DAGs),1,2 involves rules to guide the identification of variables (or combination of variables)

that must be controlled for in order to obtain unbiased effect estimates of the exposure. First, we

provide a background on the relations between the variables in our model. Second, we apply the

DAGs graphical criteria to obtain a sufficient set of variables for confounding adjustment. 1,2

Figure 3.D. Directed acyclic graphs for sufficient confounding, before (3.D.1) and after (3.D.2) the

backdoor test for sufficiency

165

Age at arrival determines both age at delivery (women aged 30 and more at arrival can only

deliver after that age) and duration of residence, defined as time from arrival to delivery (the

older the women at arrival the shorter the fertile window to be a mother). Maternal age at

delivery equals maternal age at arrival plus time from arrival to delivery. Therefore, age at

delivery is influenced directly by age at arrival, and partially via length of residence. Parity is

influenced by the previous three variables (the older the mother the more likely she is to have

had a previous birth). Age at arrival is not directly associated with preterm birth but age at

delivery, duration of residence, and parity are. Age at arrival also influences knowledge of

official languages (English/French), maternal education and marital status, because these three

variables were measured at arrival (older women were more likely to be married, to have learnt

some official language, and to bring higher educational credentials than those who arrived

younger). These three variables are also associated both with the exposure (see Table 3.1 in

Chapter 3) and the outcome (knowledge of official languages is positively associated with

preterm birth, maternal education is negatively associated with preterm birth, as well as being

married or cohabiting with their partners). Immigrant class is associated with the outcome

(refugees are at slightly higher risk of preterm birth than the rest) and also with the exposure

(refugees are more common as duration of residence increases, indicating a decline in the

refugee influx over time). Immigrant class is subsequently influenced by country of birth

(refugees originate in specific countries and the composition of family members may also vary

across countries). In the same vein, country of birth is also associated with duration of residence,

as a result of uneven waves of migration in the last two decades. Country of birth is associated

with preterm birth and also with neighbourhood material deprivation (immigrants from poor

countries tend to settle and live in poor neighbourhoods and vice versa). Thus, neighbourhood

material deprivation mediates the effects of country of origin on preterm birth. Finally, infant sex

166

is associated with preterm birth (males are at a higher risk) but is not associated with any other

variable in the graph.

Given the DAG in Figure 3.D.1 and a set S of variables in the graph that are not effects of the

exposure (duration of residence) or the outcome (preterm birth), S is sufficient for adjustment if,

upon adjustment for S, there is no unblocked backdoor path from the exposure to the outcome.

An unblocked path is a sequence of arrows connecting any two variables, irrespective of the

direction of the arrows. A path is blocked when the sequence of arrows connecting two variables

contains a variable used for adjustment (that is, included in S). In a DAG, there are only two

possible kinds of unblocked paths between variables: directed paths (direct causal connection)

and backdoor paths through a shared ancestor (that is, through a common cause of both

variables). The sequence must not contain a collider, which is represented by a variable with two

or more single arrows pointing at it (that is, a collider is a common effect of two or more

variables – a collider is a child that shares ancestors). Including a collider in S creates an

association between their common causes, even if they were not associated originally, and

therefore given a pair of variables sharing a child, if the child is included in S then one of these

two parents must be also included in S. These conditions can be checked by means of an

algorithm called the “backdoor test for sufficiency”. The test involves the following steps:

1. Delete all arrows originating from the exposure (the dashed arrows in Figure 3.D.1)

2. Draw undirected arcs to connect every pair of variables that share a child that is either in

S or has a descendant in S (parity is the only child shared by age at arrival and age at

delivery and they are already connected)

3. In the new graph resulting from applying steps 1 and 2, see whether there is any

unblocked path from exposure to disease that does not pass through S. If there is not, then

S is sufficient.

167

Regarding the association between duration of residence and preterm birth, the set S (composed

of the variables enclosed in a box in Figure 3.D.2) is sufficient for confounding adjustment, as

follows.

Control for language knowledge blocks the backdoor path preterm birth – language knowledge –

age at arrival – duration of residence. Similarly, control for maternal education and marital status

blocks similar backdoor paths connecting the outcome with the exposure. Control for immigrant

class and country of birth blocks any backdoor path connecting the outcome and the exposure via

these two variables. This also blocks the path connecting the outcome and the exposure via

neighbourhood deprivation, and therefore no adjustment for this variable is necessary. This

reflects the conventional criterion that adjustment for a mediator is not appropriate. Parity is the

only collider in S. Adjustment for parity and maternal age at delivery blocks the last backdoor

paths connecting the outcome and the exposure via maternal age at arrival, which is not

necessary for confounding control. Although it is not recommended to adjust for colliders

because this creates an association between their common causes, adjusting for maternal age at

delivery without adjusting for parity would remove the influence of maternal age of delivery on

parity, and thus parity would no longer be a collider, making it necessary to control for parity to

block the backdoor path PTB – parity – age at arrival – duration of residence. Hence, adjustment

for both maternal age at delivery and parity is necessary. Adjustment for infant sex is not

necessary because there was no association between this and any other variable in the graph and

therefore no backdoor path through this variable. This also meets the conventional criteria for

confounding in that a potential confounder is associated with both the exposure and outcome.

168

Finally, neighbourhood deprivation is associated with length of residence (Table 1 of Chapter 3).

It could also be hypothesized to be associated with age, language, education, and marital status.

However, as these covariates are included in the adjusted model any backdoor path connecting

neighbourhood deprivation with duration of residence through these variables has been blocked.

References

(1) Greenland S, Pearl J, Robins JM. Causal diagrams for epidemiologic research. Epidemiology. 1999;10(1):37-

48.


169

Appendix 3.E. Using a Cohort Approach to rule out Confounding by Cohort Effects

One alternative way of ruling out confounding by cohort effects is using a “cohort approach”.

This involves the selection of a group of immigrant women who arrived early in the study period

(landed in calendar years 1985-1988 = arrived before JAN 1 1989) so they can be followed over

time and thus ascertain their birth outcomes, compared with their non-immigrants counterparts.

As birth data span up to fiscal 2006/07, the sample size is large enough to assess duration of

residence in Canada in 5-year groups with a group composed of women 15 years and more,

which has not been done before for birth outcomes. One limitation of this approach, however,

given our data, is that birth outcomes are ascertained over 19 years (from fiscal 1988/89 to

2006/07) with the inherent problems of changing coding schemes ( ICD-9 to ICD-10-CA) and

the related issue of secular trends in the selected outcomes, which are well documented in

Ontario and elsewhere for preterm birth. Other limitations are that it is not possible to assess very

preterm birth (VPTB), moderately preterm birth (MPTB), and small for gestational age (SGA),

or to control for parity before the redesign of the DAD in fiscal 2002/03.


2,361,427 Birth records matched to a mother fiscal 1988-2006

↓ Keep those infant born within 2001 CMAs

1,802,398

↓ Exclude multiple births and stillbirths

1,752,103

↓ Keep only those mothers arriving in 1985-1988 (calendar)

1,509,962

↓ Exclude live births weighting less than 500 (due to increased

registration of near viable births) and more than 6000 g (and likely

170


data errors)

1,508,266

↓ Exclude infants born to mothers that had never been registered to

OHIP up until ‘31MAR2001’ (last date in the LIDS + 3-month

waiting period). These may be migrants arriving after ‘31DEC2000’

(last date in the LIDS) or interprovincial migrants. As these do not

appear in the available LIDS data, we do not want them to be

misclassified as ‘non-immigrants’

1,399,893

↓ Exclude records with missing data on infant sex or maternal age (not

parity)

1,399,870

↓ Exclude records with missing data on immigration characteristics

1,399,178

↓ Exclude immigrants classified as “OTHER” (i.e., other than ‘family’,

‘economic, or ‘refugee’)

1,399,043 Population size for analyses (immigrants = 69,522 and Canadian-

born/long-term residents = 1,329,521)

Table 3.E.1: Immigrants, by world region, versus non-immigrants

N LBW VLBW MLBW PTB Small PTB

Non-immigrants 1,329,521 4.23 0.62 3.60 5.03 2.72

Immigrants 69,522 5.53 0.96 4.58 5.37 3.49

Central & East Europe 1,419 4.37 1.06 3.31 4.23 3.03

Latin America & Caribbean 24,084 6.76 1.36 5.41 6.62 4.55

Middle East & North Africa 2,917 4.05 0.86 3.19 4.01 2.37

East Asia/Pacific 10,073 5.52 0.54 4.98 5.45 3.21

South Asia 9,699 6.30 0.92 5.38 5.02 3.56

Sub Saharan Africa 4,259 6.97 1.64 5.33 6.43 4.51

Industrialized Countries 17,071 3.37 0.50 2.86 3.83 2.09

Please compare Table 3.E.1 with Table 3.1 of the results section (Chapter 3).

171

Table 3.E.2. Odds Ratios* (and 85% confidence intervals) for Immigrants, by world region, versus non-

immigrants LBW VLBW MLBW PTB Small PTB

All Immigrants 1.33 (1.28-1.37) 1.54 (1.42-1.67) 1.28 (1.24-1.33) 1.13 (1.10-1.17) 1.30 (1.25-1.36)

By World region

Central & East Europe 1.05 (0.81-1.35) 1.72 (1.03-2.86) 0.93 (0.69-1.24) 0.89 (0.69-1.15) 1.14 (0.84-1.54)

Latin America & Caribbean 1.61 (1.53-1.69) 2.16 (1.93-2.42) 1.50 (1.41-1.58) 1.37 (1.30-1.45) 1.69 (1.59-1.79)

Middle East & North Africa 0.96 (0.80-1.16) 1.38 (0.93-2.05) 0.89 (0.72-1.10) 0.83 (0.69-1.00) 0.88 (0.69-1.12)

East Asia & Pacific 1.31 (1.20-1.43) 0.83 (0.64-1.09) 1.39 (1.27-1.53) 1.12 (1.03-1.23) 1.17 (1.05-1.31)

South Asia 1.56 (1.44-1.69) 1.53 (1.24-1.89) 1.55 (1.42-1.70) 1.09 (1.00-1.20) 1.37 (1.23-1.53)

Sub Saharan Africa 1.72 (1.53-1.94) 2.67 (2.11-3.39) 1.53 (1.34-1.75) 1.37 (1.21-1.55) 1.72 (1.48-1.99)

Industrialized Countries 0.80 (0.74-0.87) 0.82 (0.66-1.02) 0.80 (0.73-0.88) 0.81 (0.75-0.87) 0.78 (0.71-0.87)

* adjusted for maternal age, infant sex and fiscal year

Table 3.E.3. Number and percentage of low birthweight (LBW), preterm birth (PRT), and small for

gestational age (SGA), among residents of Ontario Census Metropolitan Areas, non-immigrants and

immigrants (arrived 1985-1988) by duration of residence in Canada, (births 1988/89 to 2006/2007) Immigrants by duration of residence

outcome Non-imm All imm < 5 y 5 – 9 y 10 -14 y 15 + y p-trend*

N 1,329,521 69,522 19,703 24,195 15,227 10,397

LBW % 4.23 5.53 4.99 5.07 6.25 6.61 <.0001

VLBW 0.62 0.96 0.75 0.90 1.12 1.24 <.0001

MLBW 3.60 4.58 4.24 4.17 5.12 5.37 <.0001

PTB % 5.03 5.37 4.21 4.69 6.34 7.74 <.0001

Small PTB 2.72 3.49 2.91 3.12 4.14 4.46 <.0001

* Cochran-Armitage Trend test (2-sided)

Please compare Table 3.E.3 with Table 3.1 of Chapter 3.

172

Table 3.E.4. Odds Ratios* (and 95% CI) comparing immigrants by duration of residence versus non-

immigrants

Immigrants by duration of residence versus non-immigrants*

Outcome <5 y 5 – 9 y 10 -14 y 15 + y

LBW % 1.20 (1.12-1.28) 1.23 (1.16-1.30) 1.49 (1.39-1.59) 1.55 (1.43-1.68)

VLBW % 1.27 (1.07-1.49) 1.48 (1.30-1.70) 1.73 (1.49-2.02) 1.84 (1.54-2.19)

MLBW % 1.18 (1.10-1.27) 1.18 (1.10-1.26) 1.43 (1.33-1.54) 1.48 (1.36-1.62)

PTB % 1.04 (0.97-1.11) 1.04 (0.98-1.10) 1.22 (1.14-1.30) 1.28 (1.19-1.38)

Small PTB% 1.13 (1.04-1.23) 1.19 (1.11-1.28) 1.50 (1.38-1.62) 1.55 (1.41-1.70)

*adjusted for infant sex, maternal age, and fiscal year.

Table 3.E.5. Odds Ratios* (and 95% CI) comparing immigrants by duration of residence

Immigrants by duration of residence*

Outcome <5 y 5 – 9 y 10 -14 y 15 + y p-trend

LBW % 1.00 1.04 (0.94-1.15) 1.29 (1.11-1.50) 1.39 (1.12-1.73) 0.0011

VLBW % 1.00 1.11 (0.87-1.42) 1.26 (0.88-1.80) 1.30 (0.78-2.15) 0.2587

MLBW % 1.00 1.03 (0.92-1.15) 1.29 (1.10-1.52) 1.41 (1.11-1.78) 0.0022

PTB % 1.00 1.03 (0.92-1.15) 1.24 (1.06-1.45) 1.38 (1.11-1.72) 0.0028

Small PTB% 1.00 1.07 (0.94-1.22) 1.37 (1.14-1.66) 1.47 (1.13-1.92) 0.0016

*adjusted for infant sex, maternal age, immigrant class, region of birth, language knowledge, high school graduation, unmarried

status, and fiscal year.



Because of the cohort approach, we also adjusted for year of birth to take account of the secular

increases in preterm birth over the last two decades.

Despite some differences in the methodology (different population, different comparison group,

changes in coding, and no adjustment for parity) these results are quite close to the main results

and support the same conclusions, particularly regarding the effects of duration of residence,

which are the focus here.

173

Appendix 4.A. Cross classified random effects model (CCREM)

The unconditional CCREM can be written as follows:

Yijk = θ000 + b0j0 + c00k + eijk

b0j0 ~ N(0, τb002), c00k ~ N(0, τc00

2), eijk ~ N(0, σ2) (equation 1)

Where Yijk is the birth weight of infant i whose mother lives in neighborhood j and comes from

country k

θ000 is the overall mean birth weight for all infants

b0j0 is the random main effect of neighborhood j, that is, the contribution of neighborhood j

averaged over all countries

c00k is the random main effect of country k

eijk is the deviation of infant ijk’s birth weight from the cell mean; that is, the difference between

the birth weight of infant i and the mean birth weight of the infants born to mothers living in

neighborhood j and immigrating from country k.

Note that if one of the random effects b0j0 or c00k is equal to zero then they drop out of the

equation and the cross-classified model reduces to a standard two-level model. The combined

model with predictors (fixed-effects) can be written as follows:

Yijk = θ000 + Xβi + Xβj + Xβk + b0j0 + c00k + eijk ,

eijk ~ N(0, σ2) (equation 2)

where Xβi is a vector of predictors at the individual-level,

Xβj is a vector of predictors at the neighborhood-level,

and Xβk is a vector of predictors at the country-level.

174

Copyright Acknowledgements

Chapter 4 “The Interplay between Immigrants’ Country of Birth and Neighbourhood Deprivation

on Birth Outcomes” is prepublication version of the article “Neighborhood Context and Infant

Birthweight Among Recent Immigrant Mothers: A Multilevel Analysis”, published in the

American Journal of Public Health, February 2009, Vol 99, No. 2, 285-293.

Permission granted by The American Public Health Association (APHA) to use the material in

this thesis is enclosed below.

Documents

Birth Outcomes of Immigrants to Urban Ontario. A ... · protocols since my Masters programs, providing useful tips to improve my theses, and my Departmental reviewers Dionne Gesink