Exploring Maternal Birthplace and Child Gender Disparities in Markers of Neglect and Maltreatment Among Young Children of Immigrants

Exploring maternal birthplace and child gender disparities in markers of neglect and maltreatment among young children of immigrants

Ariel Pulver

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy in Epidemiology Dalla Lana School of Public Health University of Toronto

Exploring maternal birthplace and child gender disparities in markers of neglect and maltreatment among young children of immigrants

Ariel Pulver

Doctor of Philosophy in Epidemiology

Dalla Lana School of Public Health University of Toronto

2019

Abstract Significant immigration to Canada has brought many cultures and parenting practices together, highlighting differences in child health and wellbeing. Given the increasingly large amount of migration from countries with high gender inequality, it is unknown how parental immigration interacts with child gender to affect healthcare and wellbeing in early childhood. In this dissertation, I present four studies regarding variation in routine preventive health care and maltreatment in very early childhood by maternal birthplace and child gender.

The first study is a scoping review where I mapped the use of gender-based analysis in research on the health of children in immigrant families. I found that child gender is an understudied aspect of immigrant children’s health, thereby presenting an opportunity for further research.

Next, in three population-based retrospective cohort studies, I compared the risk of three markers of child health care and well-being across immigrant maternal birthplaces in comparison to mothers born in Canada—immunizations, well-child visits at 24 months, and early child maltreatment at five years of age. To explore whether son preference affects child routine preventive care and maltreatment, I also compared outcomes between daughters and sons within families.

I demonstrate that children of immigrants are well cared for concerning routine immunizations and are less likely to experience maltreatment in early childhood than children of non- immigrants. Maternal birthplaces associated with high levels of gender inequity do not seem to place daughters at risk of adverse outcomes compared to sons, except for a select case. Results support addressing vaccine hesitancy and child maltreatment in the general population to promote well-being in early childhood, as well as select targeted approaches among specific immigrant groups. My studies provide a model (including data sources, study design, and analytic techniques) to monitor and detect gender inequality in the general population as well as among minority groups. My research adds to the evidence around gender equity, which will hopefully ensure girls continue to achieve the same level of health care and well-being as boys.

iii

Table of Contents Abstract ...... ii List of Tables ...... vii List of Figures ...... viii List of Appendices ...... ix Dissertation Overview ...... x Dissertation Layout ...... xi Chapter 1 Literature Review and Rationale ...... 1 1.1 Global gender inequity and the phenomenon of missing girls ...... 1 1.2 Global gender disparities in child healthcare and well-being ...... 2 1.3 Sex Ratios among Immigrant Populations ...... 5 1.4 Gender disparities in young immigrant children’s health ...... 6 1.5 Markers of child health care and well-being ...... 6 1.5.1 Immunizations ...... 7 1.5.2 Well-child visits ...... 7 1.5.3 Child abuse and maltreatment ...... 8 1.6 Health and well-being among children of immigrants ...... 9 1.7 Terminology ...... 11 1.8 Gender-based Analysis ...... 13 1.9 Rationale and Objectives of Dissertation Studies ...... 13 1.10 My role in this work ...... 17 1.11 References ...... 18 Chapter 2, Study 1: A Scoping Review of Female Disadvantage in Health Care Use Among Very Young Children of Immigrant Families ...... 25 2.1 Abstract ...... 26 2.2 Introduction ...... 27 2.2.1 Gender inequity in children’s health around the world ...... 27 2.2.2 Gender inequity in immigrant children’s health ...... 28 2.2.3 Theoretical Framework ...... 29 2.3 Methods...... 30 2.3.1 Search Strategy ...... 30 2.3.2 Screening & Data Management ...... 31 2.3.3 Quality Assessment ...... 33 2.4 Results ...... 34

2.4.1 Characteristics of the Studies ...... 34 2.4.2 Study Populations ...... 35 2.4.3 Consideration of Immigrant Status ...... 36 2.4.4 Consideration of Gender and the Use of Statistical Comparisons ...... 37 2.4.5 Health Care Utilization Outcomes ...... 37 2.4.6 Gender-Specific Reports of Study Outcomes ...... 38 2.4.7 Quality Assessment ...... 40 2.5 Discussion ...... 41 2.5.1 Importance of Gender-Based Analysis ...... 42 2.5.2 Methodological Considerations for Health Care Research of Immigrant Children ...... 43 2.5.3 Strengths and Limitations of the Study ...... 46 2.5.4 Conclusion ...... 49 2.6 References ...... 51 Chapter 3, Study 2: Differences in the receipt of infant preventive primary care among Ontario-born children across varied maternal countries of birth ...... 64 3.1 Abstract ...... 65 3.2 Introduction ...... 66 3.3 Methods...... 67 3.3.1 Overview ...... 67 3.3.2 Study Population ...... 67 3.3.3 Data Sources ...... 68 3.3.4 Variables ...... 69 3.3.5 Analyses ...... 71 3.4 Results ...... 72 3.4.1 Demographic characteristics ...... 72 3.4.2 Immunization ...... 73 3.4.3 Well-child visits ...... 73 3.4.4 Sensitivity Analyses ...... 75 3.5 Discussion ...... 75 3.5.1 Summary of findings ...... 75 3.5.2 Interpretation ...... 76 3.5.3 Strengths ...... 77 3.5.4 Limitations ...... 77 3.5.5 Implications ...... 78

3.6 Appendices ...... 86 3.7 References ...... 98 Chapter 4, Study 3: Comparison of receipt of routine preventive care among infant daughters and sons of immigrant mothers to Ontario, Canada ...... 102 4.1 Abstract ...... 103 4.2 Introduction ...... 105 4.3 Methods...... 106 4.3.1 Data sources ...... 106 4.3.2 Study population ...... 107 4.3.3 Variables ...... 107 4.3.4 Exposures ...... 108 4.3.5 Covariates ...... 109 4.3.6 Analysis ...... 110 4.4 Results ...... 111 4.4.1 Demographic characteristics ...... 111 4.4.2 Within-family gender disparities in routine preventive care ...... 111 4.4.3 Effect modification by mother tongue among infants of Indian-born mothers ...... 112 4.4.4 Results of sensitivity analyses ...... 112 4.5 Discussion ...... 112 4.5.1 Summary of findings and interpretation ...... 112 4.5.2 Strengths ...... 114 4.5.3 Limitations ...... 114 4.5.4 Implications ...... 115 4.6 Appendices ...... 120 4.7 References ...... 122 Chapter 5, Study 4: Differences in Early Childhood Maltreatment by Maternal Birthplace and Child Gender ...... 125 5.1 Abstract ...... 126 5.2 Introduction ...... 128 5.3 Methods...... 129 5.3.1 Design, participants, and setting ...... 129 5.3.2 Data sources ...... 129 5.3.3 Study variables ...... 130 5.3.4 Statistical analysis ...... 132

5.4 Results ...... 133 5.4.1 Descriptive Characteristics ...... 133 5.4.2 Differences in Early Childhood Maltreatment Rate by Maternal Region of Birth ...... 133 5.5 Discussion ...... 134 5.5.1 Conclusion ...... 137 5.6 References ...... 138 5.7 Appendices ...... 147 Chapter 6 Discussion ...... 148 6.1 Summary of findings ...... 148 6.2 Research Contributions ...... 150 6.3 Implications for Future Research ...... 152 6.4 Clinical and Public Health Implications ...... 155 6.4.1 Ensuring adequate preventive routine care ...... 155 6.4.2 Ensuring gender equity in child routine healthcare ...... 157 6.5 Methodological Considerations ...... 158 6.5.1 The use of administrative healthcare data for outcome measurement ...... 158 6.5.2 Unmeasured and Residual Confounding ...... 160 6.5.3 Generalizability ...... 161 6.6 Critical learnings for the doctoral candidate ...... 162 6.7 Conclusions ...... 163 6.8 References ...... 165

List of Tables Table 2-1 Characteristics of included studies ...... 60 Table 2-2 Description of included studies on acute, primary and general health care use ...... 61 Table 2-3 Quality assessment tool for evaluating included studies of gender disparities in immigrant children’s health care use ...... 63 Table 3-1. Descriptive demographic characteristics of mother-infant sets among the 15 maternal countries of birth with the greatest share of live deliveries in Ontario, April 2002- March 2013 ...... 80 Table 4-1 Descriptive characteristics of included mother-infant sibling sets with at least one boy and one girl, among the 15 countries with the greatest share of births in Ontario, April 2002- March 2013 ...... 116 Table 4-2 Within-family Gender Disparities in Routine Preventive Care Outcomes Stratified by Maternal Country of Birth, among all eligible siblings sets in Ontario 2002-2013 ...... 117

vii

Table 5-1 Descriptive characteristics of all eligible children to born non-immigrant and immigrant mothers in Ontario, Canada, 2002-2011 (n=1,240,946). All data are presented as a number (%) unless otherwise specified ...... 142 Table 5-2 Five-year rates and rate ratios for health-care identified early childhood maltreatment according to maternal birthplace among children born in Ontario Canada between 2002-2011...... 143 Table 5-3 Five-year rates and rate ratios for sub-components of health-care identified early childhood maltreatment identified comparing children of immigrants to children of non-immigrants born 2002-2011 ...... 144 Table 5-4 Within-family female:male odds ratios for five-year rates of health care identified early childhood maltreatment, by maternal birthplace, among all eligible children with an opposite-gender sibling born in Ontario, Canada between 2002-2011...... 145

List of Figures Figure 1-1 The life cycle of gender discrimination and health in the South Asian context [18] ...... 2

Figure 1-2 WHO regional estimates for 2008 for assessing under-5 mortality for males and females [18] . 3

Figure 2-1 Flow chart of the study selection for the scoping review of gender disparities in the health care use of very young immigrant children ...... 59

Figure 3-1 Flow Chart of Cohort Formation ...... 81

Figure 3-2 Unadjusted ■ and adjusted ● mean difference in the number of vaccine doses among Ontario- born children by age 2 years across maternal countries of birth, compared to Canadian-born mothers. Adjustment was for maternal age, birth year, neighborhood income quintile, birth order, child gender, and rurality...... 82

Figure 3-3 Unadjusted ■ and adjusted ● prevalence ratios of under-immunization among Ontario-born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers...... 83

Figure 3-4 Unadjusted ■ and adjusted ● mean difference in the number of well-child visits among Ontario-born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers...... 84

Figure 3-5 Unadjusted ■ and adjusted ● prevalence ratios of inadequate well-child visits among Ontario- born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers.. 85

Figure 4-1 Flow Chart of Cohort Formation ...... 118

Figure 5-1 Flow Chart of the Primary Study Cohort and Sibling Sub-Cohort ...... 146

viii

List of Appendices Appendix 3-1 Distribution of exclusion criteria across maternal countries of birth ...... 86 Appendix 3-2 Comparison of mother-infant sets included in the cohort to those excluded due to no documented immunizations or well-child care ...... 87 Appendix 3-3 Measurement of Well-Child Visits ...... 88 Appendix 3-4 Measurement of Immunizations ...... 89 Appendix 3-5 Decision tree for counting immunization billing codes ...... 91 Appendix 3-6 Handling outliers in the number of immunizations ...... 92 Appendix 3-7 Distribution of the number well-child visits and immunizations by infant fiscal year of birth, April 2002- March 2014 ...... 93 Appendix 3-8 Differences in under-immunization by maternal country of birth, 2002-2013 using three methods to estimate robust standard errors ...... 94 Appendix 3-9 Sensitivity Analyses Differences in Routine Care Outcomes by Maternal Country of Birth, Restricted Birth Years ...... 95 Appendix 3-10 Under-immunization using uncorrected data from all eligible births 2002-2013 ...... 96 Appendix 3-11 Differences in the number of immunizations and prevalence of under immunization among infants at 12 months old by maternal country of birth ...... 97 Appendix 4-1 Number and proportion of excluded infant siblings by maternal country of birth and gender ...... 120 Appendix 4-2 Sensitivity Analyses: Odds of routine preventive care for Ontario-born daughters compared to sons across levels of gender inequality in the maternal country of birth ...... 121 Appendix 5-1 The method used to identify child maltreatment ...... 147

Dissertation Overview The premise of my doctoral research is that parental birthplace is associated with early childhood routine preventive health care and wellbeing in a high immigration context. Parental child- rearing decisions and behaviors associated with the norms of their birthplace might influence accessing health care for their children and proclivity towards or against child maltreatment.

One understudied area is the contribution of child gender to differences in the health and well- being of children in immigrant families. Parental beliefs about gender equality may influence parental health care decision making on behalf of their children. Political, historical, religious factors influence the global variation in the social positioning of males and females. Many immigrants to Canada are born in countries where son preference is commonplace, and females are therefore subject to benign or severe forms of neglect, including inadequate routine health care and elevated early under-five mortality.

By investigating patterns of routine health care and maltreatment for girls and boys of immigrants in early childhood, we may: 1) sharpen our understanding of the health of children in immigrant families and 2) consider how gender inequality around the world may influence gender inequality within a high immigration context. Therefore, in this dissertation, I pursue a series of inquiries to identify differences in routine preventive health care and maltreatment driven by maternal birthplace and child gender. My research can help provide guidelines for future endeavours in gender equality assessment. I also hope that my research will inform relevant strategies that ensure equitable healthcare and wellbeing for children of immigrants in

Canada.

Dissertation Layout This dissertation consists of 6 chapters: Chapter 1 - Introduction; Chapter 2 – Study 1: A scoping review of female disadvantage in health care use among very young children of immigrant families; Chapter 3 - Study 2: Differences in the receipt of infant preventive primary care among

Ontario-born children across varied maternal countries of birth; Chapter 4 – Study 3:

Comparison of receipt of routine preventive care among infant daughters and sons of immigrant mothers to Ontario; Chapter 5 - Study 4: Differences in early childhood maltreatment by maternal birthplace and child gender; Chapter 6 - Discussion. Chapters 3 through 5 all have supplementary information including detailed variable operationalization and results of sensitivity analyses where applicable. The concluding discussion chapter outlines my learnings from my doctoral research program, some study constraints, research contributions, as well as practical implications and recommendations for future research.

Chapter 1 Literature Review and Rationale 1.1 Global gender inequity and the phenomenon of missing girls Given similar nutrition and medical care, females enjoy greater survival at all ages compared to males [1, 2]. However, despite a biological advantage for females, there remains excess female morbidity and mortality in many world regions. Researchers estimate that by 2035 there will be over 150 million missing women worldwide, partially attributable to excess female mortality occurring before the age of 5 [3].

The principal explanation is a deeply rooted preference for sons in many countries. Son preference is a multifactorial phenomenon which scholars attribute to resilient male lineage kinship systems, underdevelopment in educational and market opportunities for females, dowry inflation, general declines in fertility and family size, and unintended consequences of strict one- child policies [4-10]. Son preference creates male advantage within parenting practices and the household allocation of resources [4].

The consequences of son preference have been documented as early as the prenatal period. The availability of prenatal sex determination from ultrasound and genetic testing has been associated with skewed sex ratios at birth in favor of males in numerous Asian countries, including China,

India, Pakistan, Bangladesh, and South Korea [3, 4]. Research suggests this is partly due to the selective termination of female fetuses [11-14]. Male to female birth ratios even increase at higher birth orders due to mounting pressure to give birth to a boy [13, 15]. One study from India found a male to female birth ratio of 1.88 for last-born children compared to the country average of 1.07 [13]. Ultrasound to detect fetal sex has even been advertised and promoted as a method to prevent against female infanticide, decrease family size, and control the size of the population

[16]. There have been several laws designed to prevent the decline of sex ratios in India over the last 50 years but have been difficult to enforce and have not been effective [17].

1.2 Global gender disparities in child healthcare and well-being Evidence of son preference originates from South Asia and China-based research, which are dominant source regions of immigrants to Canada. Particularly in South Asia, son preference seems to continue postnatally. Through infancy and early childhood differential health investments towards sons and daughters have been documented, a phenomenon that scholars have referred to as ‘benign neglect’ or ‘health-care neglect’—see figure 1. [18, 19].

Figure 1-1 The life cycle of gender discrimination and health in the South Asian context [18]

Health care-neglect likely contributes to outlying sex ratios in infancy in North Africa, Southern

Asia, and Central Asia, and outlying sex ratios under-5 years in South and East Asia [19, 20]

(See Figure 1-2 below). For example, in India, the under-5 mortality rate is 90.2 per 1000 live births for females compared to 82.2 per 1000 for males, and the disparity persists across socioeconomic categories [21]. Since the 1990s, researchers have sought more proximal

associations with excess under-5 mortality for females, focusing on gender-based differences in

‘health-care neglect,’ by looking at patterns of health care utilization.

Figure 1-2 WHO regional estimates for 2008 for assessing under-5 mortality for males and females [18]

One recent study showed that among children under 12 years old with congenital heart disease, only 44% of families elected to have cardiac surgery for their daughters compared to 70% for their sons. The qualitative component of this study described perceived weaker matrimonial prospects, a lack of extended family support, and an unwillingness to dedicate time and resources to daughters [22]. Another study of childhood cancer patients in several Indian cities showed gender-inequitable health care seeking to the disadvantage of girls [23]. A third analysis used verbal autopsy reports in Delhi to demonstrate that female infants had more than twice the odds of death from preventable illnesses (e.g., diarrheal illness) whereas researchers observed no gender differences for less preventable illnesses (e.g., congenital anomalies, septicemia, birth asphyxia) [24]. In that same study, female infants had 3.48 times the odds of unexplained deaths compared to boys (95% CI: 1.46, 7.31) [24]. Recent evidence from Vietnam showed that after controlling for illness severity and sociodemographic characteristics, twice as many boys were

brought into the hospital emergency department than girls [25]. Moreover, parents are less likely to spend financial resources on health care for daughters than sons [26].

Female disadvantage is also well-documented concerning routine preventive community- delivered health care and basic nutrition. Girls in parts of South Asia have both lower vaccination rates and lower odds of postnatal check-up visits than boys the same age [27]. Girls are more likely to be severely stunted in both India and Pakistan, and further studies suggest this is due to inadequate breastfeeding and other nutrition [15, 28-31]. Discrimination against girls in routine health care and infant feeding seems to be greatest in families with no sons, and most dramatically affects children of higher birth order [28]. Gender bias against females has also been observed in Bangladesh [32], Nepal [33, 34] and South Korea [4]. Older studies situated in

Egypt [35] and Tunisia [36] have likewise reported lower family expenditures, immunization and primary medical care for girls compared to boys. Female disadvantage appears to vary regionally within countries. For example, states in southern India do not exemplify the same degree of gender bias as seen in northern regions [19, 37]. It is unknown whether girls are more likely to experience more severe forms of neglect, including child maltreatment and abuse.

While there is extensive literature documenting gender bias in a global context, limited research exists on gender bias in child health care and well-being among immigrant populations.

However, there is mounting evidence of gender bias among immigrant populations to high- income countries with regards to family planning, as follows.

1.3 Sex Ratios among Immigrant Populations A growing number of studies are pointing to prenatal sex selection as a problem in some immigrant communities, suggesting that son preference endures post-migration and influences family planning. Higher male:female birth ratios have been observed among immigrants from

India, South Korea, China, and Pakistan, to the USA, Canada, Australia, and the United

Kingdom, and increase at higher birth orders due to mounting pressure to deliver a son [38-43].

Findings in Norway are inconsistent [44, 45].

Sex selection in Canada has specifically been highlighted in the Punjabi immigrant population by several researchers [39, 41, 42, 46], reflecting similar patterns seen geographically in northern

India [37]. Both survey and qualitative data in the US and Canada [9, 46] show that Punjabi diasporic communities widely acknowledge the existence of sex selection. Study participants described several underlying concerns, including expectations for sons to care for elderly parents, social difficulties for women with no sons, and greater parenting challenges with daughters because girls carry the burden of honor to their family [9, 46].

There are unique contextual factors of the immigrant experience that may sustain or even enhance gender bias against females. For example, Srinivasan (2018) pointed out that many immigrants turn inward to their community for support and strength, given they may be facing marginalization in their host country [46]. The resultant bolstering of cultural identities may result in the overemphasis of gender roles rendering girls and women more vulnerable [46].

Moreover, many Punjabi women come to Canada under the family class category rather than for employment in the economic class, which may reinforce women’s vulnerability due to financial dependence [47]. Recent research has described how Punjabi immigrants are influenced simultaneously by cultural norms in India, as well as life in Canada. Srinivasan (2018) describes 5

this link between ‘back home’ and the diaspora as the manifestation of ‘transnational ties,’ which allow for patriarchal traditions to exert influence on son preference and daughter aversion [46].

1.4 Gender disparities in young immigrant children’s health While fetal sex selection among immigrants has drawn the attention of researchers and media, less is known about how gender bias may place girls in immigrant families at risk for healthcare neglect in early life. Parent-held gender biases are especially important when one considers the presiding role parents hold over their children's health and well-being at very young ages. While financial or structural barriers may impede parental health care access for their children, a parent’s ability to ensure their child’s adequate care and well-being should not differ for male or female children in the absence of gender bias. For this reason, gender differences in child healthcare and well-being before children enter school likely suggest intrahousehold gender biases. Only one small study has explicitly examined gender inequities in parental health investments for immigrant children. Researchers found no statistically significant sex differences in immunization rates of British South Asian children at 24 months, however the study had low statistical power [48] and was not designed to study gender bias within a family.

1.5 Markers of child health care and well-being To see whether ‘healthcare neglect’ (i.e., ‘benign neglect’) [18, 19] disproportionately affects girls in immigrant families, we can look to important indicators of adequate primary care— routine immunizations and well-child check-ups. Routine primary care health care visits to family physicians or pediatricians are particularly important moments in early life. These visits are useful to monitor the infant and child’s growth and developmental well-being and are moments where children receive necessary vaccinations to prevent communicable diseases.

1.5.1 Immunizations From 2 months until 18 months of age, children in Ontario are recommended to begin a series of publicly funded immunizations, which are recommended by the National Advisory Committee on Immunization and funded by the Ministry of Health and Long-Term Care. While recommendations change, they may include various iterations and combinations of antigens for diphtheria, pertussis, tetanus, polio, Haemophilus influenzae b, measles, mumps, rubella, varicella, and pneumococcal diseases. There are several ways that immunization coverage is measured at the population level and they may vary across jurisdictions. However, the 2nd birthday is a commonly used age marker to measure receipt of the full series of childhood vaccines and is thus useful for national and international comparisons [49]. In Canada, standards for ‘up-to-date’ immunization coverage require consideration of the timeliness and completeness of antigen-specific vaccine doses. ‘Up-to-date’ immunization coverage is therefore helpful to understand the level of population protection [49]. This type of coverage requires information on each specific antigen administered, the delivery date of the vaccine, and the child’s age.

An alternate, more flexible lens for immunization coverage allows for the observation of childhood vaccination behavioral patterns. By counting the number of vaccine doses against the recommended and publicly funded number, we may draw inferences about parental vaccination decision-making and variation in vaccination behaviors in different groups. Studies two and three in this dissertation make use of the alternate approach to measure infant vaccination patterns in the population.

1.5.2 Well-child visits Well-baby or well-child visits (a.k.a. check-ups) are an additional and equally important component of routine preventive care in infancy and early childhood. The Canadian Pediatric 7

Society recommends five or six routine well-baby visits from 2 to 24 months of age [50]. At these visits, children are weighed, measured, and monitored for developmental and language milestones. As well, physicians gather information about eating and sleeping. Well-child visits also provide opportunities to administer vaccinations. However, vaccination-only visits may occur, and well-child visits may occur without vaccinations. Both immunizations and well-child visits are necessary elements of health care service use, and inadequate amounts of either may indicate health-care neglect.

1.5.3 Child abuse and maltreatment A much more sinister and direct manifestation of harm to a child is the occurrence of child abuse and maltreatment. Parents or guardians perpetrate over 80% of child maltreatment, which can include physical, sexual, or psychological abuse, neglect, or involve a child witness to intimate- partner violence [51]. Child maltreatment in epidemiologic research is most often measured using data obtained from child protective services records, police services, adolescent surveys, or more recently, in administrative healthcare data. Administrative health care data provides information on child emergency room visits and hospitalizations, where child maltreatment is determined to be the cause of the injury. Researchers using healthcare administrative data to measure child maltreatment internationally have created four categories in descending order of specificity for identifying maltreatment events in the healthcare system [52]. This definition has facilitated international comparisons [52]. Categories include: 1. maltreatment syndrome

(physical abuse or neglect as the cause of injury); 2. assault (violence by caregivers (physical abuse), or others, which may be due to inadequate supervision (neglect)); 3. undetermined cause

(explicit uncertainty about the cause of injury, which may reflect physical abuse or neglect); and

4. adverse social circumstances (physician concern about parenting, home environment, or 8

broader welfare concerns associated with the injury). Marker injuries of physical abuse resulting in the child’s hospitalization can include intracranial injuries, retinal hemorrhage, and long bone or metaphyseal fractures, among pre-mobile infants without predisposing conditions (e.g., pathological fracture, osteogenesis imperfecta, birth trauma or motor vehicle accidents) [52-55].

Maltreatment ascertained from healthcare records will underestimate true prevalence as definitive investigation around causes of suspicious injuries are not always concluded prior to leaving the hospital and many cases of suspected maltreatment are not associated with physical injuries or requiring hospital care.

1.6 Health and well-being among children of immigrants There is publicly administered universal health care in Canada. The principles of universal health care include but are not limited to universality, a single-payer, and equitable access to care for all residents [56]. Therefore, in Canada, every child should have equal access to health care, including routine preventive care and acute care services. One of the purported benefits of universal health care is the amelioration of social gradients in healthcare services access [56, 57].

However, it is well documented that many social inequalities in health care service use remain

[57, 58]. Canada differs from other high immigrant-receiving countries, like the United States, due to its highly selective immigration process. In Canada, economic migrants (which comprise

60% of new Canadians) are admitted based on a points system that rewards characteristics such as age, language skills and experience, educational level, Canadian education, transferability of skills, family with Canadian citizenship, or an offer of employment in Canada [59]. The remaining 40% is mostly comprised of immigrants sponsored by Canadian family (around 25-

30%), protected persons and refugees (around 15-20%), and others (around 1%) [59].

Despite advantageous social characteristics, children in immigrant families may be at a disadvantage for receiving some routine health care services. Although there has been a considerable effort towards improving health care access for immigrants, many barriers to routine care remain [60-62]. Limited host language proficiency, low socioeconomic position, and migrating to a nascent immigrant community have been repeatedly documented as significant barriers for immigrants in accessing health care services [60, 61, 63]. Immigrants may experience greater difficulty finding physicians who take on new patients, especially among those recently arrived [64]. Lack of transportation, child care, and social support networks may be particularly challenging factors for immigrant parents, which may impact the ability to seek health care services [64]. These barriers accessing health care faced by mothers born outside of

Canada may similarly impede the accessing of health care for their young children. Health service norms in countries of birth may hinder the ascertainment of routine anticipatory care for children [65-66] and increase stigmatization by health care providers [63].

The evidence is mixed regarding the uptake of immunization among children of immigrants as compared to non-immigrants [66-70]. Limited research suggests inequalities in infant routine care to the disadvantage of immigrant children [70]. Although child maltreatment is a global phenomenon [71-73], there is disproportional representation of some ethno-racial groups in child-welfare systems compared to others [74-78]. While there is variation in maltreatment across countries, patterns are unclear as it varies by the type of maltreatment and system of measurement. Understanding culturally based parenting differences in health care and child protective services has gained importance in ethnoculturally diverse settings [74-84]. However, prior studies of child maltreatment have not distinguished between ethnic groups, have been limited to one specific group, or have combined recent immigrants with third-generation 10

families. Furthermore, few studies have explored the contribution of parental immigration to childhood maltreatment, and existing studies relied on self-reported maltreatment data [81, 85].

Gender differences in the experience of maltreatment are not entirely clear. Girls have a higher risk of being sexually abused than do boys, although rates of other types of maltreatment are similar for both sexes in high-income countries. In low-income countries, girls are at higher risk for infanticide, sexual abuse, and neglect, whereas boys seem to be at greater risk of harsh physical punishment [51]. There are no studies on gender differences in maltreatment outcomes among immigrant populations.

1.7 Terminology I will now highlight some key terminology used throughout my dissertation studies. The health of immigrant children is an area of much research focus, pointing to issues faced by children in new countries, including but not limited to their migration experience and language difficulties.

However, I would like to distinguish the experience of children who immigrate from the experience of children of immigrants who may be born in the host country. The research projects in this dissertation will focus on the experience of children of immigrants who are born in

Canada, whose parents immigrated to Canada before having their child. Therefore, the reader should take care when extrapolating the research findings to immigrant children themselves. A second point of terminology is the utilization of the words sex and gender. Biological sex of the infant is an essential variable in this dissertation since I hypothesize that child outcomes will differ between males and females born to mothers immigrating from countries where gender disparities, mainly female disadvantage, are marked. However, in this dissertation, I have chosen to use the terminology of gender, rather than sex, because it is more accurate in describing the

phenomenon of gender-based discrimination, or son preference. The value placed on male gender in various cultures influences behaviors and household practices, which is hypothesized to result in health differentials between male and female children. Finally, I would like to define the terminology of disparities, inequities, and differences, that I use throughout the dissertation to describe between-group differences (or inequalities) both in the literature and in my observations. Health disparities may contain an ethnical judgment of fairness, but not always—a disparity can be evaluated in terms of both an inequality (judgement of fairness is absent) and inequity (judgement of fairness is present), since what is unequal is not necessarily inequitable.

Margaret Whitehead’s seminal 1991 paper describes both differences that are unavoidable and fair as well as differences that are avoidable and fair [90].

Unavoidable and fair differences, or inequalities, may be due to:

1) Natural biological variation;

2) Health damaging behaviour that is freely chosen;

3) Transient health advantage of one group over another when one group is the first to

adopt a health-promoting behavior and the other group can catch up soon.

Avoidable and unfair differences, or inequities, may be due to:

1) Health-damaging behaviour when choice is severely restricted;

2) Exposure to unhealthy, stressful living and working conditions;

3) Inadequate access to essential health services and other basic services;

4) Natural selection or health-related social mobility involving the tendency for sick

people to move down the social scale.

The disparities I estimated across studies do not all necessarily represent inequities, although some might. For example, Braveman describes that while we expect young adults to be healthier than older adults, differences in immunization levels between boys and girls, or racial/ethnic differences in health treatments cause concern from an equity perspective [86].

Therefore, in my dissertation I use the following terms to denote these meanings:

1) Differences: observed differences between groups, without judgment of fairness;

2) Disparities: observed differences between groups, with or without judgement of fairness;

3) Inequities: observed differences between groups, with judgment of fairness.

1.8 Gender-based Analysis Three of the four studies in this dissertation directly use a gender-based analytic approach.

Gender-based analysis helps to recognize and clarify the differences between males and females.

It enables us to uncover inequities in health and interactions with healthcare systems [87].

Gender-based analytic approaches are critical in advancing our understanding of how social factors including race, ethnicity, and immigration status intersect with gender to shape health

[88]. The reporting of outcomes by gender is central to gender-based analysis [89].

1.9 Rationale and Objectives of Dissertation Studies Due to cultural and systemic disadvantages, girls in many world regions receive suboptimal preventive care and are more likely than boys to suffer from forms of benign and severe neglect

[4, 18, 19]. There is emerging evidence suggesting that son preference persists among some

immigrant groups particularly from countries like India, where gender inequity is high [9, 38-42,

46]. It remains unclear the extent to which son preference influences the health and well-being of young children in high immigrant-receiving settings with universal health insurance, like

Ontario. In Ontario, in the absence of gender bias, health care utilization and child maltreatment should not differ for boys and girls within a family. Immigration is the top contributor to population growth in Canada, so children born into immigrant families are a rapidly growing population—we must identify and address their unique needs.

The overall aim of my dissertation is to examine gender differences in suboptimal preventive health care and maltreatment in early childhood according to maternal birthplace and child gender. To do this, for each child health care or well-being indicator, I first examine differences by maternal birthplace and then I examine differences between siblings of opposite gender, within strata of maternal birthplace. As the proportion of children born to immigrant mothers from areas with considerable female disadvantage grows, it is crucial to study patterns in child health and well-being to identify differences for children of immigrants and non-immigrants, as well identifying any existing gender differences. Study results can help stimulate discussion to promote equitable health care use for children of immigrants and non-immigrants, for both boys and girls.

1.9.1.1 Objective 1: Using a gender-based analytic framework, map the literature of health care use for young children in immigrant families to Western, high-income countries. Rationale

Parent-held gender biases are especially important when one considers the presiding role parents hold over their child’s health care at young ages. Gender disparities may indeed continue into

infancy and early childhood in high-income, high-immigrant receiving countries, the extent to which remains largely unexplored. In infancy and preschool ages, a child’s health care use is greatly dependent on the decisions of their parents. Doctor’s visits, vaccinations, or the decision to go to the emergency department relies on parents’ actions for the benefit of their child. Should the parental-held gender biases seen in source countries of immigrants continue post-migration, one would hypothesize that some immigrant girls may receive fewer health care services compared to immigrant boys in the host country as well. While gender differences within source countries of immigrants are well documented, to date, the use of gender-based analysis in immigrant children's health research is unknown. It is important to review the literature to document if and how researchers have considered the contribution of child gender to immigrant children’s health care use (Dissertation Chapter 2).

1.9.1.2 Objective 2: Identify patterns of infant routine preventive care by 1) maternal birthplace and 2) gender within maternal birthplace groups. Rationale

In Canada, all legal residents have universal access to most primary and acute healthcare services. For children, this includes routine anticipatory primary care including well-child visits and the recommended series of vaccinations, without cost to the parent. However, even in a universal health coverage environment like Canada, immigrants face several barriers accessing health care services [62, 64]. Therefore, children in immigrant families may be at a disadvantage for receiving some routine health care services, and girls may be at a further disadvantage. To my knowledge, only one study has examined gender differences in immunization rates, but unfortunately was underpowered and not explicitly designed to identify son preference [48].

Therefore, when studying differences in routine preventive care for children of immigrants versus non-immigrants, it is also essential to disaggregate by child gender.

Canada provides an ideal setting in which to conduct health research on the children of immigrants. The most populous immigrant groups in Canada originate from countries with varied levels of gender inequality. As well, insurance status does not limit preventive health care use among children of immigrants. To address this dissertation objective, I have conducted two research studies (Dissertation Chapters 3 and 4), the first of which explores differences in primary care by maternal country of birth, and the second which explores son preference across maternal country of birth strata. As a secondary objective, because son preference has been identified among Punjabi Ontarians [39], I also explored effect modification by mother tongue among Indian-born mothers.

1.9.1.3 Objective 3: To examine patterns of early childhood maltreatment identified in the health care system by 1) maternal birthplace and 2) gender within maternal birthplace groups. Rationale

Attention has been drawn to rapidly changing immigration trends bringing together different child-rearing practices, including the perpetration of child maltreatment [77]. There is growing documentation of differences in child protective services involvement by ethnicity, race, culture, and immigration. However, few studies have explored the specific contribution of immigration to variation in early childhood maltreatment. Studies exploring immigration and maltreatment have been limited to a specific group and have relied on self-report among older children. Gender may play an important role in immigration differences in child maltreatment, associated with variation in the societal position of females around the world. Therefore, gender inequality in

maternal birthplace may influence the occurrence of early childhood maltreatment for girls and boys, in the context of immigration (Dissertation Chapter 5).

1.10 My role in this work My doctoral research program falls under an overarching research grant focused on gender disparities across the life course among immigrants in Canada. My dissertation research objectives focus on unpacking differences in markers of early childhood health and well-being by maternal birthplace and gender. With the support of my supervisor, Dr. Marcelo Urquia, I developed and refined the research questions, study populations, and study outcomes. I selected the best analytic approaches, conducted all statistical analyses, and prepared each manuscript for the studies included in this dissertation.

1.11 References 1. Hesketh, T. and Z.W. Xing, Abnormal sex ratios in human populations: causes and consequences. Proceedings of the National Academy of Sciences, 2006. 103(36): p. 13271-13275. 2. Sawyer, C.C., Child mortality estimation: estimating sex differences in childhood mortality since the 1970s. PLoS Medicine, 2012. 9(8): p. e1001287. 3. Bongaarts, J. and C.Z. Guilmoto, How many more missing women? Excess female mortality and prenatal sex selection, 1970–2050. Population and Development Review, 2015. 41(2): p. 241-269. 4. Das Gupta, M., et al., Why is son preference so persistent in East and South Asia? A cross-country study of China, India and the Republic of Korea. The Journal of Development Studies, 2003. 40(2): p. 153-187. 5. Donato, K.M., S.M. Kanaiaupuni, and M. Stainback, Sex differences in child health: Effects of Mexico-US migration. Journal of Comparative Family Studies, 2003: p. 455- 477. 6. Qian, N., Missing women and the price of tea in China: The effect of sex-specific earnings on sex imbalance. The Quarterly Journal of Economics, 2008. 123(3): p. 1251- 1285. 7. Ebenstein, A., The “missing girls” of China and the unintended consequences of the one child policy. Journal of Human Resources, 2010. 45(1): p. 87-115. 8. Ebenstein, A., H. Li, and L. Meng, The impact of ultrasound technology on the status of women in China, in Hebrew University and Tsinghua University Working Paper. 2010. 9. Puri, S., et al., “There is such a thing as too many daughters, but not too many sons”: A qualitative study of son preference and fetal sex selection among Indian immigrants in the United States. Social science & medicine, 2011. 72(7): p. 1169-1176. 10. Jayachandran, S., The roots of gender inequality in developing countries. Economics, 2015. 7(1): p. 63-88. 11. Sen, A., Missing women. BMJ: British Medical Journal, 1992. 304(6827): p. 587. 12. Sen, A., Missing women—revisited: reduction in female mortality has been counterbalanced by sex selective abortions. BMJ: British Medical Journal, 2003. 327(7427): p. 1297. 13. Arnold, F., S. Kishor, and T. Roy, Sexselective abortions in India. Population and development review, 2002. 28(4): p. 759-785. 14. Bhalotra, S. and T. Cochrane, Where have all the young girls gone? Identification of sex selection in India. 2010.

15. Barcellos, S.H., L.S. Carvalho, and A. Lleras-Muney, Child gender and parental investments in India: Are boys and girls treated differently? American Economic Journal: Applied Economics, 2014. 6(1): p. 157-89. 16. Goodkind, D., On substituting sex preference strategies in East Asia: Does prenatal sex selection reduce postnatal discrimination? Population and Development Review, 1996: p. 111-125. 17. Sahni, M., et al., Missing girls in India: infanticide, feticide and made-to-order pregnancies? Insights from hospital-based sex-ratio-at-birth over the last century. PloS one, 2008. 3(5): p. e2224. 18. Fikree, F.F. and O. Pasha, Role of gender in health disparity: the South Asian context. BMJ: British Medical Journal, 2004. 328(7443): p. 823. 19. Khera, R., et al., Gender bias in child care and child health: global patterns. Archives of disease in childhood, 2013: p. archdischild-2013-303889. 20. Alkema, L., et al., National, regional, and global sex ratios of infant, child, and under-5 mortality and identification of countries with outlying ratios: a systematic assessment. The Lancet Global Health, 2014. 2(9): p. e521-e530. 21. Collaborators, M.D.S., Causes of neonatal and child mortality in India: a nationally representative mortality survey. The Lancet, 2010. 376(9755): p. 1853-1860. 22. Ramakrishnan, S., et al., Gender differences in the utilisation of surgery for congenital heart disease in India. Heart, 2011. 97(23): p. 1920-1925. 23. Arora, R., T. Eden, and G. Kapoor, Epidemiology of childhood cancer in India. Indian journal of cancer, 2009. 46(4): p. 264. 24. Khanna, R., et al., Community based retrospective study of sex in infant mortality in India. Bmj, 2003. 327(7407): p. 126. 25. Treleaven, E., et al., Gender disparities in child health care seeking in northern Vietnam. Asian Population Studies, 2016. 12(3): p. 312-330. 26. Asfaw, A., F. Lamanna, and S. Klasen, Gender gap in parents' financing strategy for hospitalization of their children: evidence from India. Health economics, 2010. 19(3): p. 265-279. 27. Vilms, R.J., et al., Gender inequities in curative and preventive health care use among infants in Bihar, India. Journal of global health, 2017. 7(2). 28. Pande, R.P., Selective gender differences in childhood nutrition and immunization in rural India: the role of siblings. Demography, 2003. 40(3): p. 395-418. 29. Mishra, V., T.K. Roy, and R.D. Retherford, Sex differentials in childhood feeding, health care, and nutritional status in India. Population and development review, 2004. 30(2): p. 269-295.

30. Baig-Ansari, N., et al., Child's gender and household food insecurity are associated with stunting among young Pakistani children residing in urban squatter settlements. Food and Nutrition Bulletin, 2006. 27(2): p. 114-127. 31. Jayachandran, S. and I. Kuziemko, Why do mothers breastfeed girls less than boys? Evidence and implications for child health in India. The Quarterly journal of economics, 2011. 126(3): p. 1485-1538. 32. Mitra, A.K., M.M. Rahman, and G.J. Fuchs, Risk factors and gender differentials for death among children hospitalized with diarrhoea in Bangladesh. Journal of Health, Population and Nutrition, 2000: p. 151-156. 33. Pokhrel, S. and R. Sauerborn, Household decision-making on child health care in developing countries: the case of Nepal. Health Policy and Planning, 2004. 19(4): p. 218- 233. 34. Pokhrel, S., et al., Gender role and child health care utilization in Nepal. Health policy, 2005. 74(1): p. 100-109. 35. Yount, K.M., Gender bias in the allocation of curative health care in Minia, Egypt. Population Research and Policy Review, 2003. 22(3): p. 267-299. 36. Obermeyer, C.M. and R. Cardenas, Son preference and differential treatment in Morocco and Tunisia. Studies in family planning, 1997: p. 235-244. 37. Jha, P., et al., Trends in selective abortions of girls in India: analysis of nationally representative birth histories from 1990 to 2005 and census data from 1991 to 2011. The Lancet, 2011. 377(9781): p. 1921-1928. 38. Dubuc, S. and D. Coleman, An Increase in the Sex Ratio of Births to India‐born Mothers in England and Wales: Evidence for Sex‐Selective Abortion. Population and Development Review, 2007. 33(2): p. 383-400. 39. Brar, A., et al., Sex Ratios at Birth Among Indian Immigrant Subgroups According to Time Spent in Canada. Journal of Obstetrics and Gynaecology Canada, 2017. 39(6): p. 459-464. e2. 40. Ray, J.G., D.A. Henry, and M.L. Urquia, Sex ratios among Canadian liveborn infants of mothers from different countries. Canadian Medical Association Journal, 2012: p. cmaj. 120165. 41. Urquia, M.L., et al., Sex ratios at birth after induced abortion. Canadian Medical Association Journal, 2016. 188(9): p. E181-E190. 42. Urquia, M.L., et al., Variations in male-female infant ratios among births to Canadian- and Indian-born mothers, 1990-2011: a population-based register study. CMAJ Open, 2016. 4(2): p. E116. 43. Edvardsson, K., et al., Male-biased sex ratios in Australian migrant populations: a population-based study of 1 191 250 births 1999-2015. International journal of epidemiology, 2018: p. 1-13. 20

44. Singh, N., et al., Different sex ratios of children born to Indian and Pakistani immigrants in Norway. BMC pregnancy and childbirth, 2010. 10(1): p. 40. 45. Tønnessen, M., V. Aalandslid, and T. Skjerpen, Changing trend? Sex ratios of children born to Indian immigrants in Norway revisited. BMC pregnancy and childbirth, 2013. 13(1): p. 170. 46. Srinivasan, S., Transnationally relocated? Sex selection among Punjabis in Canada. Canadian Journal of Development Studies/Revue canadienne d'études du développement, 2018: p. 1-18. 47. Walton‐Roberts, M., Transnational migration theory in population geography: gendered practices in networks linking Canada and India. Population, space and place, 2004. 10(5): p. 361-373. 48. Martineau, A., M. White, and R. Bhopal, No sex differences in immunisation rates of British south Asian children: the effect of migration? BMJ, 1997. 314(7081): p. 642. 49. Canadian Immunization Research Network. National Standards for Immunization Coverage Assessment: Recommendations from the Canadian Immunization Registry Network, in Canadian Communicable Disease Report. 2012. 50. Canadian Pediatric Society. Schedule of well-child visits 2016 [cited 2018 May 30 2018]; Available from: https://www.caringforkids.cps.ca/handouts/schedule_of_well_child_visits. 51. Gilbert, R., et al., Burden and consequences of child maltreatment in high-income countries. The lancet, 2009. 373(9657): p. 68-81. 52. Gilbert, R., et al., Child maltreatment: variation in trends and policies in six developed countries. The Lancet, 2012. 379(9817): p. 758-772. 53. Warrington, S., C. Wright, and A.S. Team, Accidents and resulting injuries in premobile infants: data from the ALSPAC study. Archives of disease in childhood, 2001. 85(2): p. 104-107. 54. Kemp, A.M., et al., Patterns of skeletal fractures in child abuse: systematic review. Bmj, 2008. 337: p. a1518. 55. Schnitzer, P.G., et al., Identification of ICD codes suggestive of child maltreatment. Child abuse & neglect, 2011. 35(1): p. 3-17. 56. Naylor, C.D., Canada as Single-Payer Exemplar for Universal Health Care in the United States: A Borderline Option. Jama, 2018. 319(1): p. 17-18. 57. Martin, D., et al., Canada's universal health-care system: achieving its potential. The Lancet, 2018. 58. Bryant, T., C. Leaver, and J. Dunn, Unmet healthcare need, gender, and health inequalities in Canada. Health policy, 2009. 91(1): p. 24-32.

59. Ferrer, A.M., G. Picot, and W.C. Riddell, New directions in immigration policy: Canada's evolving approach to the selection of economic immigrants. International Migration Review, 2014. 48(3): p. 846-867. 60. Derose, K.P., J.J. Escarce, and N. Lurie, Immigrants and health care: sources of vulnerability. Health affairs, 2007. 26(5): p. 1258-1268. 61. Flores, G., M. Abreu, and S.C. Tomany-Korman, Limited English proficiency, primary language at home, and disparities in children's health care: how language barriers are measured matters. Public health reports, 2005. 120(4): p. 418-430. 62. McKeary, M. and B. Newbold, Barriers to care: The challenges for Canadian refugees and their health care providers. Journal of Refugee Studies, 2010. 23(4): p. 523-545. 63. Pottie, K., et al., Language proficiency, gender and self-reported health: an analysis of the first two waves of the longitudinal survey of immigrants to Canada. Canadian Journal of Public Health/Revue Canadienne de Sante'e Publique, 2008: p. 505-510. 64. Asanin, J. and K. Wilson, “I spent nine years looking for a doctor”: exploring access to health care among immigrants in Mississauga, Ontario, Canada. Social science & medicine, 2008. 66(6): p. 1271-1283. 65. Brown, K.F., et al., Factors underlying parental decisions about combination childhood vaccinations including MMR: a systematic review. Vaccine, 2010. 28(26): p. 4235-4248. 66. Borràs, E., et al., Parental knowledge of paediatric vaccination. BMC Public Health, 2009. 9(1): p. 154. 67. Guttmann, A., et al., Immunization coverage among young children of urban immigrant mothers: findings from a universal health care system. Ambulatory Pediatrics, 2008. 8(3): p. 205-209. 68. Borras, E., et al., Vaccination coverage in indigenous and immigrant children under 3 years of age in Catalonia (Spain). Vaccine, 2007. 25(16): p. 3240-3243. 69. Varan, A.K., et al., Vaccination Coverage Disparities Between Foreign-Born and US- Born Children Aged 19–35 Months, United States, 2010–2012. Journal of immigrant and minority health, 2017. 19(4): p. 779-789. 70. Markkula, N., et al., Use of health services among international migrant children–a systematic review. Globalization and health, 2018. 14(1): p. 52. 71. Stoltenborgh, M., et al., The prevalence of child maltreatment across the globe: Review of a series of meta‐analyses. Child Abuse Review, 2015. 24(1): p. 37-50. 72. Stoltenborgh, M., et al., Cultural–geographical differences in the occurrence of child physical abuse? A meta‐analysis of global prevalence. International Journal of Psychology, 2013. 48(2): p. 81-94. 73. Stoltenborgh, M., et al., A global perspective on child sexual abuse: meta-analysis of prevalence around the world. Child maltreatment, 2011. 16(2): p. 79-101.

74. Drake, B., S.M. Lee, and M. Jonson-Reid, Race and child maltreatment reporting: Are Blacks overrepresented? Children and youth services review, 2009. 31(3): p. 309-316. 75. Clarke, J., The challenges of child welfare involvement for Afro-Caribbean families in Toronto. Children and youth services review, 2011. 33(2): p. 274-283. 76. Kim, H., D. Chenot, and J. Ji, Racial/ethnic disparity in child welfare systems: A longitudinal study utilizing the Disparity Index (DI). Children and Youth Services Review, 2011. 33(7): p. 1234-1244. 77. Nadan, Y., J.C. Spilsbury, and J.E. Korbin, Culture and context in understanding child maltreatment: Contributions of intersectionality and neighborhood-based research. Child abuse & neglect, 2015. 41: p. 40-48. 78. Lee, B., et al., Delineating disproportionality and disparity of Asian-Canadian versus White-Canadian families in the child welfare system. Children and Youth Services Review, 2016. 70: p. 383-393. 79. Korbin, J.E., Culture and child maltreatment: Cultural competence and beyond. Child Abuse & Neglect, 2002. 26(6-7): p. 637-644. 80. Elliott, K. and A. Urquiza, Ethnicity, culture, and child maltreatment. Journal of Social Issues, 2006. 62(4): p. 787-809. 81. Alink, L.R., et al., Is elevated risk of child maltreatment in immigrant families associated with socioeconomic status? Evidence from three sources. International Journal of Psychology, 2013. 48(2): p. 117-127. 82. Cardoso, J.B., et al., Nativity and immigration status among Latino families involved in the child welfare system: Characteristics, risk, and maltreatment. Children and Youth Services Review, 2014. 44: p. 189-200. 83. Lee, B., W. Rha, and B. Fallon, Physical abuse among Asian families in the Canadian child welfare system. Journal of Aggression, Maltreatment & Trauma, 2014. 23(5): p. 532-551. 84. Schick, M., et al., Child maltreatment and migration: a population-based study among immigrant and native adolescents in Switzerland. Child maltreatment, 2016. 21(1): p. 3- 15. 85. Euser, E.M., et al., Elevated child maltreatment rates in immigrant families and the role of socioeconomic differences. Child Maltreatment, 2011. 16(1): p. 63-73. 86. Braveman, P. and S. Gruskin, Defining equity in health. Journal of Epidemiology & Community Health, 2003. 57(4): p. 254-258. 87. Clow, B., et al., Rising to the challenge: Sex-and gender-based analysis for health planning, policy and research in Canada. Halifax, NS: Atlantic Centre of Excellence for Women’s Health, 2009.

88. Greyson, D.L., A.R. Becu, and S.G. Morgan, Sex, drugs and gender roles: mapping the use of sex and gender based analysis in pharmaceutical policy research. International journal for equity in health, 2010. 9(1): p. 26. 89. Moerman, C.J. and J. van Mens-Verhulst, Gender-sensitive epidemiological research: suggestions for a gender-sensitive approach towards problem definition, data collection and analysis in epidemiological research. Psychology, Health & Medicine, 2004. 9(1): p. 41-52. 90. Whitehead M. The concepts and principles of equity and health. Health promotion international, 1991 6(3):p. 217-28.

Chapter 2 Study 1

Chapter 2, Study 1: A Scoping Review of Female Disadvantage in Health Care Use Among Very Young Children of Immigrant Families

Reference: Pulver, A., Ramraj, C., Ray, J. G., O'Campo, P., & Urquia, M. L. (2016). A scoping review of female disadvantage in health care use among very young children of immigrant families. Social Science & Medicine, 152, 50-60.

Chapter 2 Study 1

2.1 Abstract

Preference for sons culminates in higher mortality and inadequate immunizations and health care visits for girls compared to boys in several countries. It is unknown if the negative consequences of son-preference persist among those who immigrate to Western, high-income countries. To review the literature regarding gender inequities in health care use among children of parents who migrate to Western, high-income countries, we completed a scoping literature review using

Medline, Embase, PsycINFO and Scopus databases. We identified studies reporting gender- specific health care use by children aged 5 years and younger whose parents had migrated to a

Western country. Two independent reviewers conducted data extraction and a quality assessment tool was applied to each included study. We retrieved 1547 titles, of which 103 were reviewed in detail and 12 met our inclusion criteria. Studies originated from the United States and Europe, using cross-sectional or registry-based designs. Five studies examined gender differences in health care use within immigrant groups, and only one study explored the female health disadvantage hypothesis. No consistent gender differences were observed for routine primary care visits however immunizations and prescriptions were elevated for boys. Greater use of acute health services, namely emergency department visits and hospitalizations, was observed for boys over girls in several studies. Studies did not formally complete gender-based analyses or assess for acculturation factors. Health care use among children in immigrant families may differ between boys and girls, but the reasons for why this is so are largely unexplored. Further gender- based research with attention paid to the diversity of immigrant populations may help health care providers identify children with unmet health care needs.

Chapter 2 Study 1

2.2 Introduction

Gender inequity in health continues to be a deep-rooted worldwide problem. Efforts to reduce sexism over the last century have achieved much success in several Western, high-income countries, while inequities in other world regions continue, including, but not limited to, low and middle-income settings (Inglehard & Norris, 2003). In many societies, having a son is believed to be more beneficial to a family, both financially and socially. Explanations have included a perceived lack of economic utility of daughters, marginalization of females in the workforce, expensive dowry systems in some cultures, and unintended consequences of strict one-child policies (Donato et al., 2003; Ebenstein, 2010; Puri et al., 2011).

2.2.1 Gender inequity in children’s health around the world Following the introduction of prenatal sex determination through ultrasound access in various regions of India and China, higher male to female ratios at birth have been repeatedly identified (Arnold et al., 2002; Bhalotra & Cochrane, 2010; Ebenstein et al., 2010), most likely due to selectively terminating female fetuses. Male to female birth ratios even increase at higher birth orders due to mounting pressure to give birth to a boy (Arnold et al., 2002; Barcellos et al.,

2012). One study from India found a male to female birth ratio of 1.88 for last born children compared to the country average of 1.07 (Arnold et al., 2002). Other explanations have included the non-reporting of female births, female infanticide occurring the day after birth, and excess unexplained female deaths in the first year of life (Khanna et al., 2003; Sudha & Rajan, 1999).

Differential health investments towards sons and daughters seem to continue through early childhood. For example, girls in India are not breastfed for as long as boys (Barcellos et al.,

2012; Jayachandran & Kuziemko, 2009), are less likely to be immunized adequately (Pande,

2003; Pande & Yazbeck, 2003), and receive proper nutrition (Barcellos et al., 2012; Choudhury

Chapter 2 Study 1 et al., 2000; Pande, 2003). Moreover, in regions of India and Pakistan, girls are less likely to obtain health care when illness does occur compared to boys of the same age (Hasan & Khanum,

2000; Pandey et al., 2002). A study in Delhi, India demonstrated that female infants were 2 times more likely to die from preventable illnesses (e.g. diarrheal illness) whereas no gender differences were observed for less preventable illness (e.g., congenital anomalies, septicemia, birth asphyxia) (Khanna et al., 2003). Such drastic levels of the subordination of girl children in several countries have created extremely imbalanced population sex ratios including in China

(0.94), India (0.93), and Pakistan (0.90) relative to the sex ratios seen in developed countries

(1.05) (Sen, 1992, 2003).

2.2.2 Gender inequity in immigrant children’s health The gender inequalities in infant and early child health care mentioned above occur in many of the top source countries of immigrants to Western, high-income countries (UNICEF,

2009; World Bank, 2012). Recent analyses have revealed higher male to female birth ratios among immigrants from India, South Korea and Pakistan to the United States (US), Canada, and

Norway, implying that sex-selective terminations and gender discrimination in parenting decisions endure following immigration (Almond et al., 2009; Puri et al., 2011; Ray et al., 2012;

Singh et al., 2010).

Parent-held gender biases are especially important when one considers the presiding role parents hold over their children’s health care at very young ages. Gender inequities may indeed continue into infancy and early childhood in Western countries, the extent to which remains largely unexplored. From age 0 to 5, a child’s health care use depends virtually solely on the decisions of their caregivers. Doctor visits, vaccinations, or the decision to go to the emergency department rely on parents’ actions for the benefit of their child. If the parental-held gender

Chapter 2 Study 1 biases affecting very young girls’ health care persists following migration to immigrant- receiving, Western countries from immigrant-sending countries, we hypothesize that female children in such families would be at a disadvantage compared to their male counterparts.

The health of children in immigrant families is a growing priority for health care providers and policy-makers (Mendoza & Festa, 2013; Canadian Pediatric Society, 2012;

Takanishi, 2004). There have been calls for research on the health of immigrant children that focus on the child’s social context and social position, including the child’s gender (Takanishi,

2004). It is important to understand how girls’ health care experience may differ from boys in both immigrant-sending and immigrant-receiving settings (Bharadwaj et al., 2014). Therefore, the objective of this study was to review the literature regarding gender inequities in health care use among very young children, 0 to 5 years in immigrant families in Western high-income countries.

2.2.3 Theoretical Framework This research paper follows a gender-based analytic approach. Gender-based analysis helps to recognize and clarify the differences between males and females. It enables us to uncover inequities in health and interactions with health care systems (Health Canada, 2003;

Clow et al., 2009). Gender-based analytic approaches are critical in advancing our understanding of how social factors including race, ethnicity, and immigration status intersect with gender to shape health (Greyson et al., 2010). The reporting of outcomes by gender is central to gender- based analysis. It is important to collect up-to-date information on what is known about gender differences in the field of study through systematic searches of the literature and highlight the gender differences in the determinants of the given health issue (Moerman & van Mens-Verhulst,

2004). In this review, we therefore aimed to map the literature of gender-specific reporting of

Chapter 2 Study 1 health care outcomes for very young children in immigrant families to identify: 1) patterns of health care service use for immigrant boys and girls; 2) the use of gender-based analysis in immigrant children’s health care research; and 3) gaps and challenges in researching health care inequities for immigrant children.

2.3 Methods We conducted a scoping literature review of studies from Western high-income countries describing gender differences in the health care use of very young children, 0 to 5 years, in immigrant families. Arksey and O’Malley’s methodological framework for conducting scoping reviews (Arksey & O'Malley, 2005), and relevant items from the MOOSE reporting guidelines for systematic reviews of observational studies guided our methodology (Stroup et al., 2000).

Common reasons for undertaking scoping reviews include summarizing and disseminating research findings and identifying research gaps in the existing literature (Arksey & O'Malley,

2005). Rather than being guided by a strict research question with narrow inclusion and exclusion criteria such as is the case for a systematic review, the purpose of this scoping review was to identify all relevant literature, utilizing wide definitions of terms to ensure broad coverage.

2.3.1 Search Strategy Peer Reviewed Electronic Sources

This review was conducted through systematic searches of electronic library databases including Medline, Embase, and PsycINFO. The electronic search was limited to sources published from 1980 to November 2014 and those published in English. The search strategy was developed in collaboration between author AP and a medical librarian.

Chapter 2 Study 1

Three central concepts were used in combination and guided the search:

Immigration. MeSH terms included: “Emigrants and Immigrants”, “Emigration and

Immigration”, “Refugees” and “Transients and Migrants”; keywords included “immigra*”,

“refugee*”, “newcomer*”.

Children. MeSH terms included: “Infant”, “Infant, Newborn”, “Child”, “Pediatrics”,

“Child, Preschool”; keywords included: “infan*”, “child”, “neonat*”, “newborn*”, “pediatric*”,

“paediatric*”.

Gender. MeSH terms included: “Sex Factors”, “Sex Distribution”, “Gender Identity”; keywords and phrases included “sex difference*”, “sex factors”, “sex distribution*”, “sex based difference*”, “sex based disparit*”, “gender difference*”, “gender disparit*”, “gender specific*”,

“gender equity*”, “gender inequit*”, “gender based*”, “gender related*”.

Next, hand searching of key journals, including those addressing pediatrics, maternal child health, and immigration, in addition to Scopus, Google, and Google Scholar searches were used to locate literature, including any grey literature that may have been missed through the other electronic databases from inception up until November 2014. Hand searching of the reference lists of all included literature and additional review articles was conducted for relevant missed titles. These results were subject to the same inclusion/exclusion criteria as those identified through the database searches.

2.3.2 Screening & Data Management Title Screening Retrieved titles were scanned for relevance. Titles were selected for abstract screening if they included one or more words related to the three guiding concepts— immigration, children, or gender.

Chapter 2 Study 1

Abstract Screening Abstracts were considered for inclusion if they: 1) related to children in immigrant families in Western, or otherwise industrialized settings (i.e. European Union,

United Kingdom, United States, Canada, Australia, or New Zealand); 2) focused on a health service use outcome; and 3) included very young children 0 to 5 years of age in the study sample. One author (AP) conducted the title and abstract screening in consultation with another author (MU) to discuss uncertainties with the search strategy. Two of the authors (AP and MU) met to discuss study inclusion and exclusion at the beginning of the scoping process as well as during the abstract screening.

Articles were included in the analysis if the content was relevant to health care use outcomes and the age range included but was not necessarily limited to children 0 to 5 years.

This decision was made because if strict age criteria were enforced, very few articles would be selected for review. Studies that exclusively dealt with school-aged children and adolescents were excluded. In instances where the gender-relevance of the article was unclear from the title or abstract, the article was selected for full text review. No restrictions were placed on the type of immigrant (refugee, labour migration, family reunification, etc.), the country of origin, or birthplace of the child. We chose to focus on the use of health care services rather than health outcomes more generally because health care use is more directly related to parenting decisions on behalf of children at this very young age. While child health outcomes like obesity, injury, infections, asthma etc., may be influenced by parental decision-making, their etiologies are likely more complex.

Full Text Review Full text reviews were conducted for 103 studies. Two reviewers screened the full-text articles independently (AP and CR), and discrepancies were resolved by consensus and consultation with a third author (MU). Gender was rarely part of the central

Chapter 2 Study 1 research question in the studies retrieved; therefore the reporting of gender-specific figures for immigrant children guided the full text reviews for inclusion. The full text reviews gleaned 12 studies that were eligible for inclusion.

Data Management and Extraction Endnote software was used to manage retrieved articles and Excel was used to build a database for results. Data charting methods were used in order to synthesize and analyze the results of the review. The charting step involved the documentation of key characteristics and information of the articles being reviewed.

Information was extracted on the citation, study location, study objective, general study design, study population, the immigration-related variable, how immigrant status was measured, the age range of the study sample, any confounder variables for the gender analysis, the main outcome variables, how the outcome variables were measured, the sample size, the use of gender in the analyses and the gender-relevant findings of the study. Two reviewers independently carried out the data extraction and discrepancies were resolved by consensus. It became evident that meta-analysis would be inappropriate due to the considerable heterogeneity in study settings, populations, exposures and outcomes. The charted data formed the basis for the analysis in what is known as the ‘descriptive-analytical method’ from the narrative tradition (Arksey & O'Malley,

2005), which involves collecting standard information from each research report and applying a common analytic framework to all included studies.

2.3.3 Quality Assessment The quality assessment tool was created to evaluate the appropriateness of the study to answer our question of identifying gender inequities in health care use within immigrant groups of very young children. Quality assessments are not often a component of scoping reviews and their importance is still up for debate (Levac et al., 2010). However, we found it was important

Chapter 2 Study 1 to consider study quality because of the nascent state of the literature and as a way to identify areas where more work is needed. The Newcastle-Ottawa Quality Assessment Scale for Cohort and Cross-Sectional Studies was used as a guide to inform the development of the tool (Wells et al., 2000). This scale is a very well established and validated quality assessment for conducting systematic reviews. By developing our own quality measurement, we are able to see how well the studies are equipped to answer our study question, while simultaneously assessing the scope of the literature.

The quality assessment tool was used to assess six criteria:

1. Aside from immigrant or non-immigrant, were other immigration-related factors

considered (e.g., acculturation, length of residence, language, immigrant type, country

or region of origin)?

2. Did the analysis focus on children age 0 to 5 years?

3. Was there a non-immigrant comparison group?

4. Were rates adjusted for age or other factors?

5. Was the ascertainment of exposure valid (secure record or structured interview)?

6. Was outcome assessment valid (independent blind assessment or record linkage)?

Studies received one point for each quality assessment item. Scores of 1-2 were Low Quality, 3-

4 were Medium Quality, and 5-6 were High Quality.

2.4 Results

2.4.1 Characteristics of the Studies A total of 12 peer-reviewed articles met inclusion criteria. A flow chart of the study selection process is displayed in Figure 2-1. Nine studies were conducted in Europe (United

Chapter 2 Study 1

Kingdom (1); Spain (2); Italy (2); Denmark (2); Belgium (1), and four were from the United

States.

All included studies involved empirical quantitative analyses. Several qualitative papers were excluded during the review stage because they did not meet inclusion criteria.

Characteristics of the included studies are displayed in Table 2-1. Seven studies were based on secondary data analysis of population-based administrative databases of health records (Dyhr et al., 2007; Fedeli et al., 2011; Gimeno-Feliu et al., 2009; Gimeno-Feliu et al., 2013; Laursen &

Møller, 2009; Martineau et al., 1997; Muennig et al., 2004), two were cross-sectional surveys within health care settings (e.g., emergency departments, general practitioner clinics) (Grassino et al., 2009; Van der Stuyft et al., 1993), three were cross-sectional in-person household surveys

(Berg et al., 2004; Mohanty et al., 2005; Weathers et al., 2003), with one being record-linked to national medical expenditure data (Mohanty et al., 2005).

2.4.2 Study Populations Only two studies were restricted to very young children ≤ 5 years of age (Fedeli et al.,

2011; Martineau et al., 1997), one of children 24 months and younger (Martineau et al., 1997) n=461, and the other of children ages 1-5 years old (Fedeli et al., 2011) n=29,522. Six others considered children ages 0 and up, to ages 14 through 18 (Gimeno-Feliu et al., 2009; Gimeno-

Feliu et al., 2013; Grassino et al., 2009; Laursen & Møller, 2009; Mohanty et al., 2005; Van der

Stuyft et al., 1993), one for children 0-12 years (Weathers et al., 2003), one included children 3-

17 years of age (Berg et al., 2004), and one included children 1-18 years (Dyhr et al., 2007).

Additionally, one study analyzed both children and adults together (Muennig et al., 2004).

Chapter 2 Study 1

The size of the study population falling within the 0-5 year age-range was only identifiable in Fedeli et al (2011) and Martineau et al. (1997). Therefore we cannot consistently report on the effective sample sizes for the very young age range of 0 to 5, but have reported the sample size of immigrant children in each study in Table 2-2.

2.4.3 Consideration of Immigrant Status The definition of immigrant varied across included studies (see Table 2-2). Four studies defined children’s immigrant status based on the birthplace of their parents, however they did not distinguish whether the children themselves were born in the source or receiving countries.

(Fedeli et al., 2011; Grassino et al., 2009; Laursen & Møller, 2009; Weathers et al., 2003). These studies considered children to be ‘immigrants’ if one or both or their parents were foreign-born, regardless of the child’s place of birth. This definition aligns with the current study’s hypothesis, which is that parental country of origin is the important source of variation in gender-held biases that may influence children’s health care use. Three studies did not clarify the basis of their immigrant definition, but were formed on ethnic origin (Berg et al., 2004; Martineau et al., 1997;

Van der Stuyft et al., 1993). Further, three studies comprised subjects born outside of the receiving country (Dyhr et al., 2007; Mohanty et al., 2005; Muenning et al., 2004).

The two Spanish studies considered immigrant children to be foreign nationals (Gimeno-

Feliu et al., 2009; Gimeano-Feliu et al., 2013). Three included studies further categorized immigrants by country or region of origin (Dyhr et al., 2007; Laursen & Moeller., 2009;

Grassino et al., 2009), and another (Van der Stuyft et al., 1993) collected data on acculturation- related variables (language facility, length of stay in receiving country, etc.); however these nuanced characteristics were not ascribed to any gender-specific estimates of children’s health care use. Only one study examining children’s health care utilization based its analysis explicitly

Chapter 2 Study 1 on immigrants from regions where son-preference is well documented (children of South Asian origin) (Martineau et al., 1997). A second study did focus on Indian-born immigrants, however it was not motivated by the son-preference hypothesis (Muenning et al., 2004).

2.4.4 Consideration of Gender and the Use of Statistical Comparisons In general, there were three treatments of gender across the studies. In five studies, gender was a determinant of health care use within stratified immigrant or non-immigrant groups. In those studies, direct statistical comparisons were made between genders, within immigrant groups (See Table 2-2). In four studies, immigrant status was the determinant and results were stratified by gender. Here, the statistical comparisons were made between immigrant groups within each gender. In three of the studies, gender was purely for descriptive purposes. In these studies rates of health care use by gender and immigrant status were presented graphically.

In all cases, for this review, data on gender-specific rates, proportions, frequencies or effect estimates for immigrant groups were extracted.

2.4.5 Health Care Utilization Outcomes Five studies reviewed investigated one or more acute outcomes such as hospitalizations

(Berg et al., 2004; Fedeli et al., 2011; Laursen & Møller., 2009; Muennig et al., 2004), emergency/casualty department visits (three studies) (Berg et al., 2004; Grassino et al., 2009;

Laursen & Møller, 2009) and urgent care clinics (one study) (Berg et al., 2004). Seven studies focused on primary care or general health care use outcomes including vaccination (Martineau et al., 1997), prescription drug use (Gimeno-Feliu et al., 2009), primary care services (Dyhr et al.,

2007; Gimeno-Feliu et al., 2013; Van der Stuyft et al., 1993), any examination by a physician

(Weathers et al., 2003), and health care expenditures (Mohanty et al., 2005). Hospitalizations,

Chapter 2 Study 1 emergency department visits, and primary care service use were the most frequently studied outcomes.

2.4.6 Gender-Specific Reports of Study Outcomes Primary care or general health care use

Primary care visits (Dyhr et al., 2007; Gimeno-Feliu et al., 2013; Van der Stuyft et al.,

1993) and prescription drug utilization (Gimeno-Feliu et al., 2009) were less frequently reported among immigrant girls than immigrant boys. However, no statistical comparisons were made between genders and gender-related variations appeared similar for immigrant and non- immigrant groups.

Vaccine uptake and timely immunizations may be more common among some immigrant boys than girls. One study explicitly examining son preference in the UK examined South Asian

Muslim and non-Muslim children, suggests gendered patterns for Measles, Mumps, Rubella

(100% South Asian non-Muslim boys vs. 93% South Asian non-Muslim girls), and Polio vaccinations (94% South Asian Muslim boys vs. 89% South Asian Muslim girls) compared with children of European descent where rates were similar across genders (Martineau et al., 1997).

Differences were not statistically significant. Authors attributed their null finding to a possible effect of a more egalitarian context for families new to the UK.

In contrast to the above findings, two American studies revealed opposite gender patterns. In one of the two studies that adjusted for child’s health status, female children of migrant workers 0 to 12 years of age had 2.26 times the odds of having any physician exam in the past 3 months than males (95% CI:1.28-3.98). This study also controlled for care access (e.g., transportation, availability of interpreters) and socio-demographic factors (Weathers et al., 2003).

Chapter 2 Study 1

Similarly, Mohanty and colleagues’ study of immigrants’ and non-immigrants’ health care expenditures suggest higher mean expenditures among females than males in both groups

(2005), however statistical comparisons were not made between genders. Moreover, in this study, gender-specific results were not separated for children and adults; therefore higher expenditures for females may be related to maternity care costs.

Acute Care

The majority of studies point to higher rates of acute care service use among immigrant boys compared to immigrant girls. In Berg and colleagues’ study (2004) of 193 Latino children with asthma age 3-17 years of age in San Diego, California, boys had greater odds of emergency department visits than girls (OR: 3.0, 95% CI: 1.4-6.4). However, no statistically significant differences were seen for urgent care clinics or hospitalizations. This study controlled for factors relating to health status, including asthma factors, as well as adjusting for health care coverage and socio-demographic variables (Berg et al., 2004). A registry-based study of Indian-born people in New York reported no differences in hospitalization between males and females

(Muennig et al., 2004). Data were not shown, and were not analyzed separately for children.

Gender-based comparisons were simply descriptive for the remaining three acute care studies and therefore we cannot conclude there are true differences between boys and girls within groups of immigrants without performing statistical tests between genders. In all of these studies, acute care rates were highest in immigrant boys followed by immigrant girls. One study presented results that suggested a greater gender-based difference may exist among immigrants than non-immigrants (Laursen & Møller, 2009). Authors examined differences in acute care use for injuries and reported age-adjusted rates for boys and girls of Danish-born mothers, Western

Chapter 2 Study 1 born mothers, and non-Western mothers. Gender differences were greatest for children of Non-

Western born mothers (boys: 191/1000 person-years vs. girls: 124/1000 person-years), followed by children of Western-born mothers (boys: 165/1000 person-years vs. girls: 133/1000 person- years). Gender differences were the least evident in children of Danish mothers (boys: 193/1000 person-years vs. 169/1000 person-years). Rates for hospitalization followed the same pattern

(Laursen & Møller, 2009).

A study of pediatric emergency department visits in Italy by immigrant children and

Italian-born children reported that among immigrant children, 57.8% of visits were boys and

42.2% of visits were girls, but patterns of visits across genders were similar in non-immigrants

(Grassino et al., 2009). This finding was echoed in a second Italian study of pediatric hospitalizations where crude rates of both medical and surgical related diagnoses were elevated for immigrant boys compared to immigrant girls (Fedeli et al., 2011). In this study, patterns of health care use by gender were mostly consistent across children, aside from medical-related diagnoses where the gender difference was more pronounced for Italian than for immigrant children (Fedeli et al., 2011). Only crude rates were reported and not compared between genders within immigrant groups, therefore differences in rates may be due to chance.

Table 2-2 includes descriptions of each included study and summarizes whether health care use in that study was more frequent among immigrant boys, immigrant girls, or neither, and indicates whether differences are based upon statistical comparisons or simply crude rates.

2.4.7 Quality Assessment The results of the quality assessment for each study are presented in Table 2-3. The most common issues were the reporting of only crude gender-specific rates, the heterogeneity in

Chapter 2 Study 1 children’s ages, and the heterogeneity in immigrant groups. Two studies were of low quality, six were medium and four were of high quality.

2.5 Discussion

The purpose of this review was to systematically map the literature pertaining to gender equity in the health care use of very young children in immigrant families. Three central issues were revealed through the search, data extraction, and analysis stages of this review, which will form the basis for the current discussion:

1) Immigrant girls may use health services less frequently than their male counterparts,

depending on the study setting and ethnic group.

2) Gender-based analyses of immigrant children’s health care use are lacking.

3) Several methodological challenges exist in immigrant health care research and require

consideration in studying this population.

We found that immigrant girls across study settings generally use acute care services, including emergency departments and hospitalizations, less frequently than immigrant boys. For routine care services, including primary care visits, vaccinations and medication use, results were equivocal across studies. Very young immigrant boys generally had higher rates of uptake of vaccinations than girls, however differences were not statistically significant (Martineau et al.,

1997), and authors therefore concluded, perhaps hastily, that there were no important differences across genders. The study results may still carry clinical and public health significance, as the study was designed to detect a 16% difference in immunization rates between genders, which is very large. Unfortunately, authors did not consider duration of residence in the UK, which could mediate such a relationship. Studies of primary care use did not reveal any marked gender

Chapter 2 Study 1 inequities in immigrant groups of children; however US studies revealed more physician exams and greater health expenditures among immigrant females than males. Results from primary care use align with an earlier US study suggesting that gender norms in child health investments may shift following migration (Donato et al., 2003). The finding of increased acute care use among immigrant boys is consistent with an Indian study that found that boys under nine years of age were more likely to be hospitalized than girls. In that study, parents were more willing to borrow and sell financial assets for boys’ health care costs than those for girls (Asfaw et al., 2010).

2.5.1 Importance of Gender-Based Analysis The results of this review reveal a large research gap in the area of immigrant children’s health— the use of gender-based analysis. Gender-based analysis may be germane for studying very young immigrant children’s health care use. If parents originate from societies where gender preferences for family composition and female disadvantage through the life-course exist, they may be less motivated to facilitate their children’s health care encounters. At this early life stage, children arguably spend all their time with their caregivers; as they gain increasing independence through school age and adolescence, the decision makers about their health care use may change (Alderson & Montgomery, 1996).

Clow and colleagues (2009) identify the power that gender-based analysis has in identifying who is affected by health inequities, by enabling the researcher to ask questions such as, “‘Do we have evidence about females and males of all ages or for adults or children only?” and “Do we have information about females and males from diverse ethnic and socioeconomic backgrounds?.” With respect to the health of very young children in immigrant families, the answer is no to both questions. The current review documented only 12 studies that reported gender-specific estimates health care use within immigrant children. Gender-specific reporting of

Chapter 2 Study 1 immigrant children’s health outcomes is necessary in order to elucidate where inequities in child health may remain. Moreover, rarely did studies explicitly highlight the gender differences in the origins and scope of child’s health care utilization, a crucial step for gender-sensitive research

(Moerman & van Mens-Verhulst, 2004).

Additional depth can be gained from further steps in gender-based analysis, including obtaining data from interviewing subjects about their lived experience (Moerman & van Mens-

Verhulst, 2004). We did not have the chance to fully explore the richness of this theoretical approach because the studies meeting inclusion criteria did not allow for that. The quality assessment pointed to other challenging methodological areas in conducting health research about immigrant children.

2.5.2 Methodological Considerations for Health Care Research of Immigrant Children The influence of culturally specific variation deserves further attention in the study of health care use of very young immigrant children. Finer delineation of immigrant groups by country or region of origin, ethnic, or religious membership, can lay a theoretical foundation for understanding culturally driven gender differences in children’s health care use. For example,

Donato demonstrated more favorable shifts in the health of young Mexican girls whose parents had more experience migrating to and from Mexico and the US in comparison to boys (Donato et al., 2003). It is unknown if the frequency and duration of the parental migration experience may affect the health of girls and boys of other immigrant backgrounds differently, due to unique socio-cultural environments, traditions, and practices around gender (WHO Report, 2012). Only one of the included studies hypothesized that health care among girls would be sub-optimal based on the immigrant population, which was limited to the South-Asia region where discrimination against daughters is documented (Martineau et al., 1997). Therefore the

Chapter 2 Study 1 heterogeneity of immigrant groups, which could include countries with both high and low levels of discrimination against girls, in many of the included studies is likely to mask any effect of gender on immigrant children’s health care utilization. Female disadvantage however is documented in several world regions including South and East Asia, North and Sub-Saharan

Africa, the Middle East, and Latin America (The World Bank, 2012).

Similarly, distinguishing between newcomers, longer-term immigrants and various categories of immigrants (e.g., refugee, family re-unification, labor migration) is important in future research endeavors for several reasons. First, health insurance coverage often differs substantially based on these factors (BeLue et al., 2014; Canadian Pediatric Society, 2012;

Weathers et al., 2008), and thus can predetermine a child’s access to health care services, regardless of parental-held gender-biases. Second, substantial work has identified changes in health status and behaviors of immigrants, as duration of residence in the host country increases

(De Maio, 2010; Salehi, 2010). This includes research suggesting that a girl’s health may benefit more from lengthier parental migration experience than a boy’s (Donato et al., 2003). Third, parents may originate from different countries, including from the host country, which could modify the gender-health care use relationship for some children. Variations in parental immigration factors should therefore be considered for study as potential effect modifiers.

There is evidence demonstrating that discrimination against girls in health care and infant feeding practices is greatest in families with no sons, and with children of higher birth order

(Mishra et al., 2005). Unfortunately, no studies in the current review accounted for family composition or birth order in estimating children’s health care use. We believe that future research would benefit from such lines of inquiry.

Chapter 2 Study 1

Only two of the included studies accounted for the health status of children with respect to generating effect estimates of gender on health care encounters (Berg et al., 2004; Weathers et al., 2003). In Berg and colleagues’ study (2004), adjustment for health care need among children with asthma pointed to greater emergency department use among boys than girls, but no significant differences were found for hospitalizations or urgent care clinics. In contrast,

Weather’s study (2003), revealed girls had greater odds of health care utilization when adjusted for need. Both studies adjusted for other, different factors as well, which may limit comparability.

A further area for contemplation is what different health care encounters reflect in terms of health or ill health for immigrant children. For instance, do more frequent primary care visits mean that a child is receiving the recommended schedule of care or that they require more frequent monitoring because they are unwell? Deciding to take a child to an emergency department could similarly represent a serious acute issue requiring swift care, or it could be more accessible to a family than scheduling an appointment with the child’s regular physician.

Even a large proportion of hospitalizations are discretionary (Goodman et al., 1994;

McConnochie et al., 1999). Several studies in this review found greater use of acute care services and medications in very young ages for immigrant boys than immigrant girls, which begs a related question—is excess use due to greater medical need, or greater access? In a number of the included studies, it is unclear whether the unequal use of health care services between males and females suggests inequities, an unfair and avoidable difference in health care use, or an actual gendered difference in health care need. Boys may be more risk-oriented and therefore more often require medical attention for injuries than girls (Andrade et al., 2013). However, elevated risk-taking among boys is not an innate sex characteristic, but is a result of socialization

Chapter 2 Study 1 and parenting (Booth & Nolen, 2012; Morrongiello & Hogg, 2004) and therefore it may vary across cultures with different gender expectations. Going forward, selection and interpretation of health care use outcomes to investigate inequities in immigrant children’s health care should be well defined, theory-driven and setting dependent.

A final design consideration involves the use of non-immigrant comparison groups.

While gender inequities within immigrant groups of children is the focus of the current review, situating gender inequities in immigrant children next to inequities in native-born children would provide a context to help ascertain the extent to which this effect is unique for immigrant populations rather than broader phenomenon underlying any observed differences. Few studies in this review allowed for such a comparison.

2.5.3 Strengths and Limitations of the Study To our knowledge, this is the first study to apply a gender-based approach to the study of very young immigrant children’s health care utilization. Some methodological strengths of this review include the broad and varied use of search terms in order to scope out the literature

(Arksey & O'Malley, 2005), the use of a quality assessment tool designed specifically for the included studies, and the duplicate data extraction by two reviewers. Additionally, we can have confidence in the validity of many of the exposure and outcome measures of the included studies because they frequently came from registries of secured records rather than self-report.

However, there are several limitations of this review that warrant discussion beyond the aforementioned challenges. Included studies were limited to those published in English, therefore important studies may have been missed. While two reviewers conducted the data extraction and inclusion/exclusion decisions, the author alone conducted the quality assessment.

This may have introduced a source of bias in the analysis. Moreover, the quality assessment

Chapter 2 Study 1 identified numerous issues that restrict our ability to make conclusive inferences about such patterns. A central problem of many studies was that the age range of included subjects exceeded our target age of 0 to 5 years. Most of those studies did not report age- and gender-stratified values for health care use. As the gender comparison was our study priority, we therefore frequently were only able to extract gender-specific values for the whole study sample rather than the target age group. This decision may have introduced some heterogeneity to our review, as gender biases among parents may manifest differently across childhood and adolescence, and vary in their influence on children’s health care use. This selection bias may have attenuated the observed effects across studies.

Another limitation of this review is that the method of analyzing and reporting gender differences varied meaningfully across studies. Some studies reported crude frequencies, proportions or means, while others reported rates or multivariate adjusted effect estimates.

Adjustment for confounders and health care need was inconsistent across studies. The diverse methodological nature of reported estimates inhibited the comparability of included studies as well as any meta-analyses. A number of the included studies did not conduct statistical tests to identify differences between immigrant boys’ and girls’ health care utilization. These studies however are valuable from a hypothesis-generating standpoint, and reveal patterns in health care use for boys and girls that must be confirmed with future empirical studies that make use of large sample sizes in low probability events for children such as hospitalizations.

A further hindrance is that aside from acute care services, only one study was available for several health care use outcomes including vaccinations, expenditures, and medications, highlighting the shortage of research in this area.

Chapter 2 Study 1

It is also important to consider the health care policies in the study settings and how they may affect immigrant populations’ access to health care in general. In the United States, almost half of noncitizen immigrants lack health insurance, in comparison to naturalized citizens and

US-born citizens where non-coverage is around 15-20% (Derose, Escarce, Lurie 2007). The children of immigrants are also disproportionately uninsured compared to US-born children whose parents were also born in the US (Huang & Ledsky 2001) and are less likely to have a regular source of health care, even after adjusting for insurance. In our scoping review, two of four US studies accounted for insurance coverage in their analysis comparing males and female

(Berg et al., 2004; Weathers et al., 2003). Although it is possible that going forward with the introduction of the Affordable Care Act (ACA) that greater numbers of immigrants and their children will have health insurance coverage, it would still be prudent for US studies of immigrant children’s health care use to consider insurance in their analyses. This is especially important given that children in mixed-status families (undocumented parents with US-born citizen children) may have differential health care access under the ACA (Orteaga et al., 2015).

In a similar vein, dual foreign-born couples may provide health care access to their children differently than mixed couples, and these may further vary according to whether the mother or the father is foreign-born.

Italy, Spain, the UK and Denmark all provide universal health care coverage through national health systems for documented immigrants. The provision of different services varies for undocumented immigrants in each country, but include urgent care generally, child health care until the age of 18 in Spain, preventive care and treatment for communicable diseases in

Italy, and treatment for some illnesses and family planning in England (Vázquez et al., 2011). In

Denmark, health care access for undocumented migrants is quite restricted aside from emergency

Chapter 2 Study 1 services (Jensen et al., 2011). In Belgium, immigrants may access health care services through the compulsory health insurance that covers the majority of health expenditures with patient co- pay reimbursed partially through the administration of sickness funds or through social assistance (Corens 2007; Lorant, Van Oyen & Thomas, 2008). While there may be differential access to health care for immigrants and non-immigrants due to insurance coverage, it should not necessarily create a gender imbalance within the groups themselves. However, if female disadvantage exists to a greater degree in uninsured populations than in insured ones, it might lend weight to the son-preference hypothesis, as families may be spending more resources on their sons than daughters when resources are limited.

2.5.4 Conclusion Immigrant children’s health care use differs from that of non-immigrant children

(Mendoza, 2009; Takanishi, 2004; UNICEF, 2009), and immigrant children face unique barriers to health care use, including difficulties with insurance coverage (BeLue et al., 2014), language facility (Yu & Singh, 2009), and limited socioeconomic resources (Johnson-Motoyama, 2014).

Children of immigrants are a rapidly growing proportion of Western populations (UNICEF,

2009), yet many gaps remain in the health services literature about predictors of their access to care. Studies are scarce that explore the role of child gender in limiting or facilitating timely access to health care services.

This review found some gender-based differences in the use of acute and routine care services among immigrant children, although results should be interpreted with caution as several methodological challenges were noted. Researchers studying health care of immigrant children are therefore encouraged to consider gender differences when planning their analyses.

Attention must be given to the diversity of immigrant communities in order to identify those that

Chapter 2 Study 1 may be more or less likely to hold gender-biases, including parents originating from regions where girls’ health care disadvantage has been recognized (Arnold et al., 2002; Barcellos et al.,

2012; Bhalotra & Cochrane, 2010; Choudhury et al., 2000; Ebenstein, 2010; Hasan & Khanum,

2000; Jayachandran & Kuziemko, 2009; Khanna et al., 2003; Pande, 2003; Pande & Yazbeck,

2003; Pandey et al., 2002; Sen, 1992, 2003; Sudha & Rajan, 1999). Neglecting to examine gender as an independent effect may mask a potentially important driver of health care use in groups of immigrant children. Results from future studies are important to help pediatricians and other health care providers identify families who may have children with unmet health care needs. Furthermore, results can inform interventions to promote equitable health care use across genders, immigrant groups, and ethnicities.

Chapter 2 Study 1

2.6 References Alderson, P., & Montgomery, J. (1996). Health care choices: making decisions with children:

Institute for Public Policy Research.

Almond, D., Edlund, L., & Milligan, K. (2009). O sister, where art thou? The role of son

preference and sex choice: evidence from immigrants to Canada. National Bureau of

Economic Research.

Andrade, C., Cordovil, R., & Barreiros, J. (2013). Injuries in preschool children: the hypothetical

protector effect of minor injuries and risk factors for minor and medically attended

injuries. International Journal of Injury Control and Safety Promotion, 20, 239-244.

Arksey, H., & O'Malley, L. (2005). Scoping studies: towards a methodological framework.

International Journal of Social Research Methodology, 8, 19-32.

Arnold, F., Kishor, S., & Roy, T. (2002). Sex‐Selective Abortions in India. Population and

Development Review, 28, 759-785.

Asfaw, A., Lamanna, F., & Klasen, S. (2010). Gender gap in parents' financing strategy for

hospitalization of their children: evidence from India. Health Economics, 19, 265-279.

Barcellos, S.H., Carvalho, L., & Lleras-Muney, A. (2012). Child Gender and Parental

Investments in India: Are Boys and Girls Treated Differently? National Bureau of

Economic Research.

BeLue, R., Miranda, P.Y., Elewonibi, B.R., & Hillemeier, M.M. (2014). The association of

generation status and health insurance among US children. Pediatrics, 134, 307-314.

Berg, J., Wahlgren, D.R., Hofstetter, C.R., Meltzer, S.B., Meltzer, E.O., Matt, G.E., et al. (2004).

Latino Children with Asthma: Rates and Risks for Medical Care Utilization. Journal of

Asthma, 41, 147-157.

Chapter 2 Study 1

Bhalotra, S.R., & Cochrane, T. (2010). Where have all the young girls gone? Identification of

sex selection in India. Discussion paper series//Forschungsinstitut zur Zukunft der Arbeit.

Bharadwaj, P., Dahl, G.B., & Sheth, K. (2014). Gender Discrimination in the Family. The

Economics of the Family: How the Household Affects Markets and Economic Growth [2

volumes], 237.

Booth, A.L., & Nolen, P. (2012). Gender differences in risk behaviour: does nurture matter? The

Economic Journal, 122, F56-F78.

Canadian Pediatric Society (2012). Caring for Kids New to Canada. Retrieved from

http://www.kidsnewtocanada.ca/

Choudhury, K.K., Hanifi, M.A., Rasheed, S., & Bhuiya, A. (2000). Gender inequality and severe

malnutrition among children in a remote rural area of Bangladesh. Journal of Health,

Population and Nutrition, 123-130.

Clow, B.N., Pederson, A., Haworth-Brockman, M., & Bernier, J. (2009). Rising to the challenge:

Sex-and gender-based analysis for health planning, policy and research in Canada:

Atlantic Centre of Excellence for Women's Health Halifax, NS.

Corens D. Health system review: Belgium. Health Systems in Transition, 2007; 9(2): 1–172.

De Maio, F.G. (2010). Immigration as pathogenic: a systematic review of the health of

immigrants to Canada. International Journal of Equity in Health, 9, 1-20.

Derose, K. P., Escarce, J. J., & Lurie, N. (2007). Immigrants and health care: sources of

vulnerability. Health Affairs, 26(5), 1258-1268.

Donato, K.M., Kanaiaupuni, S.M., & Stainback, M. (2003). Sex Differences in Child Health:

Effects of Mexico-US Migration. Journal of Comparative Family Studies, 34, 455-477.

Chapter 2 Study 1

Dyhr, L., Andersen, J.S., & Engholm, G. (2007). The pattern of contact with general practice and

casualty departments of immigrants and non-immigrants in Copenhagen Denmark.

Danish Medical Bulletin, 54, 226-229.

Ebenstein, A. (2010). The “missing girls” of China and the unintended consequences of the one

child policy. Journal of Human Resources, 45, 87-115.

Ebenstein, A., Li, H., & Meng, L. (2010). The Impact of Ultrasound Technology on the Status of

Women in China. Hebrew University Working Paper.

Fedeli, U., Alba, N., Zambon, F., Avossa, F., & Spolaore, P. (2011). Hospitalisation for surgical

procedures is lower among immigrants than among non-immigrant preschool children:

results from a region-wide study in Italy. Journal of Epidemiology & Community Health,

65, 645-647.

Gimeno-Feliu, L.A., Armesto-Gámez, J., MacIpe-Costa, R., & Magallán-Botaya, R. (2009).

Comparative study of paediatric prescription drug utilization between the spanish and

immigrant population. BMC Health Services Research, 9.

Gimeno-Feliu, L.A., Magallón-Botaya, R., Macipe-Costa, R.M., Luzón-Oliver, L., Cañada-

Millan, J.L., & Lasheras-Barrio, M. (2013). Differences in the use of primary care

services between Spanish national and immigrant patients. Journal of Immigrant and

Minority Health, 15, 584-590.

Goodman, D.C., Fisher, E.S., Gittelsohn, A., Chang, C.-H., & Fleming, C. (1994). Why are

children hospitalized? The role of non-clinical factors in pediatric hospitalizations.

Pediatrics, 93, 896-902.

Chapter 2 Study 1

Grassino, E.C., Guidi, C., Monzani, A., Di Pietro, P., & Bona, G. (2009). Access to paediatric

emergency departments in Italy: A comparison between immigrant and Italian patients.

Italian Journal of Pediatrics, 35.

Greyson, D.L., Becu, A.R., & Morgan, S.G. (2010). Sex, drugs and gender roles: mapping the

use of sex and gender based analysis in pharmaceutical policy research. International

Journal for Equity in Health, 2, 26.

Hasan, I.J., & Khanum, A. (2000). Health care utilization during terminal child illness in squatter

settlements of Karachi. Journal of Pakistan Medical Association, 50.

Health Canada. (2003). Gender-Based Analysis. Retrieved from http://www.hc-sc.gc.ca/hl-

vs/gender-genre/analys/gender-sexes-eng.php

Huang, Z. J., Yu, S. M., & Ledsky, R. (2006). Health status and health service access and use

among children in US immigrant families. American Journal of Public Health, 96(4),

634-640.

Inglehart, R., & Norris, P. (2003). Rising tide: Gender equality and cultural change around the

world: Cambridge University Press.

Jayachandran, S., & Kuziemko, I. (2009). Why do mothers breastfeed girls less than boys?

Evidence and implications for child health in India. National Bureau of Economic

Research.

Jensen, N. K., Norredam, M., Draebel, T., Bogic, M., Priebe, S., & Krasnik, A. (2011). Providing

medical care for undocumented migrants in Denmark: what are the challenges for health

professionals?. BMC health services research,11(1), 154.

Chapter 2 Study 1

Johnson-Motoyama, M. (2014). Does a paradox exist in child well-being risks among foreign-

born Latinos, U.S.-born Latinos, and Whites? Findings from 50 California cities. Child

Abuse and Neglect, 38, 1061-1072.

Khanna, R., Kumar, A., Vaghela, J., Sreenivas, V., & Puliyel, J. (2003). Community based

retrospective study of sex in infant mortality in India. BMJ, 327, 126.

Laursen, B., & Møller, H. (2009). Unintentional injuries in children of Danish and foreign-born

mothers. Scandinavian Journal of Public Health, 37, 577-583.

Levac, D., Colquhoun, H., & O'Brien, K. K. (2010). Scoping studies: advancing the

methodology. Implementation Science : IS, 5, 69. http://doi.org/10.1186/1748-5908-5-69

Lorant, V., Van Oyen, H., & Thomas, I. (2008). Contextual factors and immigrants’ health

status: double jeopardy. Health & place, 14(4), 678-692.

Martineau, A., White, M., & Bhopal, R. (1997). No sex differences in immunisation rates of

British south Asian children: the effect of migration? BMJ, 314, 642-643.

McConnochie, K.M., Russo, M.J., McBride, J.T., Szilagyi, P.G., Brooks, A.-M., & Roghmann,

K.J. (1999). Socioeconomic variation in asthma hospitalization: excess utilization or

greater need? Pediatrics, 103, e75-e75.

Mendoza, F.S. (2009). Health disparities and children in immigrant families: A research agenda.

Pediatrics, 124, S187-S195.

Mendoza, F.S., & Festa, N.K. (2013). New American children: Supporting the health and well-

being of immigrant populations. JAMA Pediatrics, 167, 12-13.

Mishra, V., Roy, T. K., & Retherford, R. D. (2004). Sex differentials in childhood feeding,

health care, and nutritional status in India. Population and development review, 30(2),

269-295.

Chapter 2 Study 1

Moerman, C., & van Mens-Verhulst, J. (2004). Gender-sensitive epidemiological research:

suggestions for a gender-sensitive approach towards problem definition, data collection

and analysis in epidemiological research. Psychology, Health & Medicine, 9(1), 41-52.

Mohanty, S.A., Woolhandler, S., Himmelstein, D.U., Pati, S., Carrasquillo, O., & Bor, D.H.

(2005). Health care expenditures of immigrants in the United States: A nationally

representative analysis. American Journal of Public Health, 95, 1431-1438.

Morrongiello, B.A., & Hogg, K. (2004). Mothers' reactions to children misbehaving in ways that

can lead to injury: Implications for gender differences in children's risk taking and

injuries. Sex Roles, 50, 103-118.

Muennig, P., Jia, H., & Khan, K. (2004). Hospitalization for heart disease, stroke, and diabetes

mellitus among Indian-born persons: A small area analysis. BMC Cardiovascular

Disorders, 4.

Ortega AN, Rodriguez HP, Vargas Bustamante A. Policy dilemmas in Latino health care and

implementation of the Affordable Care Act. Annu Rev Public Health. 2015 Mar

18;36:525-44.

Pande, R.P. (2003). Selective gender differences in childhood nutrition and immunization in

rural India: the role of siblings. Demography, 40, 395-418.

Pande, R.P., & Yazbeck, A.S. (2003). What's in a country average? Wealth, gender, and regional

inequalities in immunization in India. Social Science & Medicine, 57, 2075-2088.

Pandey, A., Sengupta, P.G., Mondal, S.K., Gupta, D.N., Manna, B., Ghosh, S., et al. (2002).

Gender differences in healthcare-seeking during common illnesses in a rural community

of West Bengal, India. Journal of Health, Population and Nutrition, 306-311.

Chapter 2 Study 1

Puri, S., Adams, V., Ivey, S., & Nachtigall, R.D. (2011). "There is such a thing as too many

daughters, but not too many sons": A qualitative study of son preference and fetal sex

selection among Indian immigrants in the United States. Social Science & Medicine, 72,

1169-1176.

Ray, J.G., Henry, D.A., & Urquia, M.L. (2012). Sex ratios among Canadian liveborn infants of

mothers from different countries. Canadian Medical Association Journal. DOI:10.1503

/cmaj.120165

Salehi, R. (2010). Intersection of health, immigration, and youth: A systematic literature review.

Journal of Immigrant and Minority Health, 12, 788-797.

Sen, A. (1992). Missing women. BMJ, 304, 587.

Sen, A. (2003). Missing women—revisited. BMJ, 327, 1297-1298.

Singh, N., Pripp, A.H., Brekke, T., & Stray-Pedersen, B. (2010). Different sex ratios of children

born to Indian and Pakistani immigrants in Norway. BMC Pregnancy and Childbirth, 10,

40.

Stroup, D.F., Berlin, J.A., Morton, S.C., Olkin, I., Williamson, G.D., Rennie, D., et al. (2000).

Meta-analysis of observational studies in epidemiology: a proposal for reporting. JAMA,

283, 2008-2012.

Sudha, S., & Rajan, S.I. (1999). Female demographic disadvantage in India 1981–1991: Sex

selective abortions and female infanticide. Development and Change, 30, 585-618.

Takanishi, R. (2004). Leveling the playing field: Supporting immigrant children from birth to

eight. The Future of Children, 61-79.

UNICEF. (2009). Children in Immigrant Families in 8 Affluent Countries: Their family, national

and international context New York, NY.

Chapter 2 Study 1

Van der Stuyft, P., Woodward, M., Amstrong, J., & De Muynck, A. (1993). Uptake of

preventive health care among Mediterranean migrants in Belgium. Journal of

Epidemiology and Community Health, 47, 10-13.

Vázquez, M. L., Terraza-Núñez, R., Vargas, I., Rodríguez, D., & Lizana, T. (2011). Health

policies for migrant populations in three European countries: England; Italy and

Spain. Health Policy, 101(1), 70-78.

Weathers, A., Minkovitz, C., O'Campo, P., & Diener-West, M. (2003). Health services use by

children of migratory agricultural workers: Exploring the role of need for care.

Pediatrics, 111, 956-963.

Weathers, A., Minkovitz, C., Diener-West, M., & O'Campo, P. (2008). The effect of parental

immigration authorization on health insurance coverage for migrant Latino children.

Journal of Immigrant and Minority Health, 10, 247-254.

Wells, G., Shea, B., O’Connell, D., Peterson, J., Welch, V., Losos, M., et al. (2000). The

Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in

meta-analyses.

The World Bank. World Development Report 2012. Gender Equality and Development. 2012.

Washington, DC, The World Bank.Yu, S.M., & Singh, G.K. (2009). Household language

use and health care access, unmet need, and family impact among CSHCN. Pediatrics,

124, S414-S419.

Chapter 2 Study 1

Figure 2-1 Flow chart of the study selection for the scoping review of gender disparities in the health care use of very young immigrant children

Chapter 2 Study 1

Table 2-1 Characteristics of included studies

Study characteristics (n=12) n (%)

Design Population-based health registry 7 (58%) Cross-sectional study (hospital or clinic setting) 2 (17%)

Cross-sectional (household survey) 3 (25%)

Country

United States 4 (33%)

Western Europe 8 (67%)

Basis of immigrant definition

Foreign birthplace of parents 4 (33%)

Foreign birthplace of child 3 (25%) Ethnicity or country of origin 3 (25%) Foreign-national 2 (17%) Gender consideration Gender comparison within stratified immigrant groups 5 (42%)

Immigrant comparison within stratified gender groups 4 (33%)

Gender as descriptive variable only 3 (25%)

Chapter 2 Study 1

Table 2-2 Description of included studies on acute, primary and general health care use

Author, Year; Study Design; Data Source Immigrant Children Basis of Immigrant Outcome More Frequent Region Sample Definition Among Immigrant Boys, Girls or Neither Gimeo-Feliu et al., Population-based registry study; Immigrant children of Foreign national Defined daily Boysa 2009; Aragon, Aragonese Health Service databases Aragon, 0-14 years old; doses/1000 Spain and SALUD Pharmacy Service billing n=19,231 persons/day (DID) database Mohanty et al., Cross-sectional survey with record Immigrant children 0-18 Non-US born Per Capita health Femalesa 2005; United States linkage;1998 Medical Expenditure years old; n=276 care expenditures No data shown Panel Survey linked to 1996–1997 specific for children National Health Interview Survey Van der Stuyft et Cross-sectional study of 33 medical Immigrant children 0-14 Moroccan, Turkish or Any preventive Neitherb al., 1993; Belgium offices years old; n=661 other Mediterranean care immigrant children ethnicity

Weathers et al., Cross-sectional household survey Migrant families with Children of migrant Physician exam in Girlsc 2003; North one randomly selected agricultural workers the past 3 months Carolina, United child 0-12 years old; States n=300

Martineau et al., Population-based registry study based South Asian Muslim South Asian children % vaccine uptake: 1997; Newcastle, on Newcastle Child Health Register and non-Muslim born in the Newcastle United Kingdom children, 24 months old; registry MMR Muslim: Neitherb n=461 Non-Muslim: Boysc Tetanus Muslim: Neitherb Non-Muslim: Boysb Polio Muslim: Boysb Non-Muslim: Boysb Pertussis Muslim: Boysb Non-Muslim: Boysb Diphtheria Muslim: Neitherb Non-Muslim: Boysb BCG Muslim: Girlsb Non-Muslim: Girlsb

Chapter 2 Study 1

Table 2-2 continued. Description of included studies on acute, primary and general health care use

Author, Year; Region Immigrant Children Sample Basis of Immigrant Outcome More Frequent Among Definition Immigrant Boys, Girls or Neither Dyhr et al., 2007; Immigrant children with a permanent Longer than All service use Boysa Copenhagen, Denmark address in Copenhagen accessing 3 months’ residence permit services; n= 7903 and foreign citizenship

Gimeno-Feliu et al., Immigrant children of Zaragoza 0-15 Foreign national Total number of health Boysa 2013; Zaragoza, Spain years old; n=7730 centre visits per year

Berg et al., 2004.; Latino families with a child with asthma Born to a self-identified ED visits Boysc California, US 3-17 years old; n=193 Latino parent, 70% foreign- Hospitalizations Neitherb born Urgent care visits Neitherb

Fedeli et al., 2011; Immigrant children 1-5 years of age in Foreign citizenship Medical hospital Boysa Veneto, Italy Veneto; (born outside Italy or born admissions n=29,522 in Italy to foreign parents) Surgical hospital Boysa admissions

Grassino et al., 2009; Foreign pediatric patients to emergency Parental birthplace outside ED visits Boysa Italy department 0-18 years old; European Union n=2437 Laursen et al., 2009; Children of immigrants living in 32 Maternal country of birth Emergency department Boysa Denmark municipalities across Denmark visits for injury 0-15 years old; n=23,640 Hospital admissions Boysa Muenning et al., 2004; Indian-born children in New York City; Birth place in India Hospitalization chronic Neitherb New York, US N=7,489 diseases

a. Indicates that no statistical testing was performed and greater frequency between genders is based on crude rates b. Indicates that statistical testing was performed and differences between genders were not statistically significant and greater frequency is based on crude rates c. Indicates that statistical testing was performed and differences between genders were statistically significant

Chapter 2 Study 1

Table 2-3 Quality assessment tool for evaluating included studies of gender disparities in immigrant children’s health care use

Gimeno- Gimeno- Van der Berg et Dyhr et Fedeli et Grassino Laursen Martinea Mohanty Muennin Weathe Feliu et Feliu et Stuyft et al. al. al. et al. et al. u et al. et al. g et al. rs et al. al., 2009 al., 2013 al. Study Population 1. Was immigrant status defined by a variable in 1 0 0 0 0 0 1 1 0 1 1 1 addition to being foreign born? 2. Did the analysis focus 0 1 1 1 1 0 0 1 0 0 0 0 on preschool age children? Study Design 3. Was there a non- immigrant comparison 0 1 1 1 1 1 1 1 1 1 0 1 group? Comparability 4. Were rates adjusted for 1 0 1 0 0 0 1 0 0 1 1 1 age or other factors? Measurement Validity 5. Was the ascertainment of exposure valid? (secure 0 1 1 1 1 1 1 1 1 1 0 0 record or structured interview) 6. Was outcome assessment valid? (independent blind 0 1 1 1 1 1 1 1 1 1 0 1 assessment or record linkage) Score 2 4 5 4 4 3 5 5 3 5 2 4 Study Quality Low Medium High Medium Medium Medium High High Medium High Low Medium 1-2 Low Quality; 3-4 Medium Quality; 5-6 High Quality

Chapter 3 Study 2

Chapter 3, Study 2: Differences in the receipt of infant preventive primary care among Ontario-born children across varied maternal countries of birth

Chapter 3 Study 2

3.1 Abstract Background

Barriers experienced by immigrants may impede attaining routine health care for their young children. The current study explored whether there is variation in routine immunization and well-child visits by maternal country of birth in the offspring of immigrants to Canada, where universal health care is available.

Methods

We conducted a retrospective population-based cohort study using administrative data in

Ontario, Canada. Included were 1,079,713, singleton term infants followed for 24 months, born between 2002 and 2014 to mothers from 15 different countries, including Canada. Modified

Poisson regression models estimated adjusted prevalence ratios (aPR) to compare the prevalence of under-immunization and inadequate well-child visits among infants of each maternal country to those of Canadian-born mothers. Models accounted for the possibility of more than one birth per mother, as well as other confounders.

Results

For most maternal countries of birth, infants received a greater number of routine immunizations compared to Canada. However, under-immunization was more likely among those of USA (aPR

1.06, 95% CI 1.03, 1.09), Poland (aPR 1.06, 95% CI 1.03, 1.08), and Somalia-born mothers (aPR

1.09, 95% CI 1.05, 1.12). Inadequate well-child visits were more common among 7 of the 14 foreign countries of birth, including China (aPR 1.31, 95% CI 1.28, 1.35), India (aPR 1.14, 95%

CI 1.11, 1.16), Jamaica (aPR 1.17, 95% CI 1.13, 1.22), USA (aPR 1.17, 95% CI 1.11, 1.24), Sri

Chapter 3 Study 2

Lanka (aPR 1.03, 95% CI 1.00, 1.07), Pakistan (aPR 1.33, 95% CI 1.30, 1.37) and Somalia (aPR

1.84, 95% CI 1.77, 1.92).

Conclusions

In the first two years of life, the offspring of most immigrant women receive more vaccinations, but many have fewer well-child visits than infants of Canadian-born mothers.

3.2 Introduction In Canada, all legal residents have universal access for most primary and acute healthcare services. For children, this includes routine anticipatory primary care, including well-child visits and the recommended series of vaccinations, without cost to the parent. Well-child visits represent crucial moments when children are monitored for healthy growth and development and receive necessary vaccinations against infectious diseases. Even in Canada’s universal health coverage environment, children in immigrant families may be at a disadvantage for receiving health care services. Despite considerable effort towards improving health care access, many barriers to routine care remain for new Canadians [1, 2]. In recent years, approximately 50-60% of Canada’s more than 250,000 annually admitted immigrants are admitted as economic migrants (including principal applicants, spouses and dependents), via a points system rewarding highly educated individuals and their families proficient in English or French. The remaining

40% is mostly comprised of immigrants sponsored by Canadian family (around 25-30%), protected persons and refugees (around 15-20%), and others (around 1%) [3].

Despite Canada’s uniquely selective system, healthcare barriers remain for immigrants, and particularly for refugee immigrants. Reported barriers include low socioeconomic position, limited host language proficiency, and migrating to a nascent immigrant community [1, 4, 5].

Chapter 3 Study 2

Immigrants in Canada report experiencing greater difficulty finding physicians, as well as difficulty obtaining transportation, child care, and social support [6]. Health beliefs and health service norms in countries of birth may also affect health care seeking [7, 8].

Evidence is mixed regarding the uptake of immunization among children of immigrants as compared to non-immigrants [9-12]. Research suggests inequalities in infant routine care to the disadvantage of immigrant children [12], however research is limited among varied contemporary immigrant groups. Therefore, the objective of this study was to identify whether there are differences in infant routine anticipatory care (immunization and well-child visits) across common maternal countries of birth (MCOB) in Ontario, Canada in comparison to infants of Canadian-born mothers. Ontario is an ideal research setting to achieve the research objective because of the diversity of the immigrant population and the universal health care environment where insurance status does not limit access to health care.

3.3 Methods

3.3.1 Overview We built a population-based retrospective cohort including all infants born in Ontario, Canada between April 1st 2002 and March 31st, 2014 and followed for routine health encounters for the first 24 months of life, until March 31st, 2016. We linked maternal immigration records and hospital delivery records to identify women who immigrated to Canada and delivered babies in

Ontario hospitals.

3.3.2 Study Population Inclusion/Exclusion

The study population included healthy singleton term infants born in Ontario from April 1st,

2002 to March 31st, 2014 eligible for routine preventive primary care in Ontario from a

Chapter 3 Study 2 pediatrician or family physician/general practitioner until 24 months. Included were infants whose mothers were born in 15 countries with the greatest share of Ontario births, representing over 90% of all births to immigrant mothers during the study period.

Excluded infants were those born at <37 weeks gestation or <2500 grams at birth, who were multiples, or if they were diagnosed with a complex chronic condition (e.g., cancer, major congenital malformations among others) before the age of 2 [13], as this may influence their experience of routine care.

Infants were removed from the study cohort if they had no documented well-child visits or immunizations, or if they received any primary care from community health centers (CHCs) as these physicians do not bill OHIP for their activities. Infants with no health care billings may see a salaried physician or other health care provider who does not submit billings for visits or immunizations. If infants had at least one record of care in the CHC database, they were flagged and removed from the cohort. Mothers or infants with any missing covariate or outcome data were excluded (Figure 3-1). Appendices 3-1 and 3-2 contain the distribution of exclusion factors across MCOB, and differences in demographic factors between included and excluded infants, respectively.

3.3.3 Data Sources Data for this study come from several linked population-based administrative databases at ICES.

The Registered Persons Database (RPDB) is the provincial health care registry, and a unique coded identity number facilitates record linkage between the databases. The RPDB contains information on birth date and year, sex, and postal code, which was linked to Canadian Census data to obtain neighborhood information at the level of a dissemination area.

Chapter 3 Study 2

The Discharge Abstract Database of the Canadian Institute for Health Information was used to identify all in-hospital deliveries (capturing 98% of births). Well-child visits and vaccinations with family physicians and pediatricians were captured using the Ontario Health Insurance Plan

(OHIP) claims database, which contains information on physicians’ billings, such as fee codes for visits, diagnostic codes, and date of service.

Information on maternal birthplace and date of landing to Canada was sourced from the Ontario portion of the federally maintained Immigration, Refugees, and Citizenship Canada (IRCC)

Permanent Resident Database. The IRCC database was sourced for documentation of all landed immigrants who obtained permanent residence from Jan 1, 1985, to Dec 31, 2012. Many earlier studies have linked the IRCC with the other databases [14-17]. The overall linkage rate for the

IRCC database to the RPDB is 86.4% [14-17]. Unmatched mothers were classified as non- immigrants. Immigrants might be misclassified as non-immigrants if they arrived before 1985

[17].

3.3.4 Variables

3.3.4.1 Outcomes Number of vaccinations by a family physician or pediatrician by 24 months of age. An immunization was measured by identifying physician billing codes for the administration of a vaccine and the number of units delivered on a given day. Unscheduled vaccinations occurring before six weeks did not count towards the total. Earlier studies have demonstrated high specificity and sensitivity, as well as higher coverage estimates in administrative data as compared to immunization cards, and lower estimates than parental self-report [9, 18-20].

Infants were categorized as under-immunized if they had received fewer than the expected number of vaccinations publicly funded, according to the province’s immunization schedule.

Chapter 3 Study 2

Infants born 2002-2003, 2004-2009, 2010, and 2011-2013 should have 5, 12, 11, and 10 vaccines respectively, against the availability in Canada at the time (e.g., DPTP/Hib, MMR, pneumococcus, meningococcus and varicella vaccines). The term under-immunization is meant to reflect less than the expected number of total administered vaccine doses. We did not examine specific antigens, therefore under-immunization here does not represent coverage necessarily

[21]. A small proportion of children had a clinically unrealistically high count of vaccinations.

An immunization dose count above the 99th percentile was removed or otherwise was corrected to the number publicly funded in that year + 2.

Number of well-child visits by a family doctor or pediatrician by 24 months of age. These were operationalized by a set of core primary care fee codes and diagnostic codes and excluded immunization-only visits [22]. Infants were determined to have inadequate well-child visits if they had less than the expected number of well-child visits (5 visits) with a family doctor or pediatrician (recommended at 2, 4, 6, 12, 15 (optional) and 18 months of age) in the first two years of life [23].

Vaccines or well-child visits documented two weeks after 24 months were included as a buffer for appointment scheduling. See Appendices 3-6 for further details on variable operationalization, codes, and management. As the publicly funded vaccine schedules and health care funding models have changed over the study period, the distribution of the frequency of visits is described by birth year in Appendix 3-7.

3.3.4.2 Exposures Maternal country of birth (MCOB) is recorded during the immigration application process from notarized copies of original documents and entered in the IRCC permanent resident database. If mothers did not have an immigration record, they were determined to be non-immigrants and

Chapter 3 Study 2 assumed to be Canadian-born. The proportion of misclassified immigrants is not expected to meaningfully bias estimates of the Canadian-born group. Mother-infant pairs among the 15

MCOB with the greatest share of births, including Canada, were selected for the most significant population health relevance and to ensure robust sample sizes.

3.3.4.3 Covariates Confounders were selected a priori. They included maternal age at delivery of the index child

(≤19, 20-34, ≥35 years of age), infant birth year [24], birth order (1st, 2nd, 3rd, 4th or more) [19], neighborhood income quintile (1 being the lowest and 5 being the highest), and urban/rural residence (urban=<40 on Rurality Index of Ontario; rural= ≥40) [25]. We sourced covariate data from the hospital record (maternal age, infant birth year, birth order), and Canadian census data

(neighborhood income quintile, and urban/rural residence).

Statistics Canada’s Postal Code Conversion File links the mother’s postal code at delivery to census data to generate values for the census-related variables. Infants delivered on or before

December 31st, 2003 were linked with 2001 census values, and the remaining births were linked with 2006 census values.

3.3.5 Analyses

3.3.5.1 Primary Analyses Cross-tabulations and univariate procedures were used to obtain baselines counts of immunizations and well-child visits, as well as the prevalence of under-immunization and inadequate well-child visits across MCOB. To attain adjusted estimates of differences in the

(mean) number of vaccinations and well-child visits and prevalence ratios (aPR) for each MCOB compared to Canada, we used multivariable modified Poisson regression with generalized estimating equations (GEE). GEE was used to generate robust standard error estimates (RSE) to

Chapter 3 Study 2 account for non-independence of children of the same mother as well as to estimate prevalence ratios for binary outcomes from longitudinal data [26, 27]. Models were also adjusted for infant birth year, infant gender, maternal age group, birth order, neighborhood income quintile, and urban/rural residence.

3.3.5.2 Sensitivity Analyses We ran several sets of sensitivity analyses to evaluate the possible influence of the choice of statistical methods, data quality, and consistency over time. First, we tested three methods to generate RSEs in case a given method would elicit differing results or precision estimates

(Appendix 3-8). Given the variation in immunization recommendations and the availability of

CHC data over the study period, we: i) stratified the cohort by different immunization eras

(2002-2004 and 2005-2013); and ii) restricted the cohort to 2008-2013 births where complete

CHC data were available (Appendix 3-9). Although the percentage of children whose immunization data was corrected was small, we ran analyses on the complete original data to see if the correction introduced an information bias (Appendix 3-10).

Finally, we checked if inadequate immunization using recommendations by 12 months of age would produce different results as recommendations by 24 months (Appendix 3-11), which would help generate hypothesize about potential mechanisms for future study.

3.4 Results

3.4.1 Demographic characteristics Table 3-1 describes the characteristics of mothers and infants from each included MCOB group.

The total eligible population was 1,079,713 healthy, term, singleton infants, of mothers from 15 countries with the greatest share of births in Ontario. MCOB represented included Canada, India,

Chapter 3 Study 2

China, Pakistan Philippines, Sri Lanka, Jamaica, Poland, Vietnam, Afghanistan, Guyana, USA,

Iraq, Iran, and Somalia. Somalia had the most adverse sociodemographic profile with the highest proportion of families in the lowest income quintile (69.9%), fourth or higher birth (31.4%), and the most refugees (63.9%).

3.4.2 Immunization The crude, cumulative number of immunizations ranged between 8.2 (Somalia) to 10.8

(Vietnam) (Figure 3-2). In general, under-immunization was common across all MCOB. The crude prevalence of under-immunization ranged from 27.1% for Vietnam to 43.1% for Canada, to 53.2% for Somalia (Figure 3-3).

Figure 3-2 presents regression models estimating the crude and adjusted mean differences in the number of vaccine doses and Figure 3-3 shows PRs and aPRs of under-immunization, across countries in comparison to Canada. Following covariate adjustment, infants of mothers born in

USA, Poland, and Somalia had significantly fewer immunizations than those born in Canada and were 6 to 9% more likely to be under-immunized. Infants of mothers born in all other MCOB had a significantly higher number of immunizations and were protected from under- immunization compared to children of non-immigrants.

3.4.3 Well-child visits The crude cumulative number of well-child visits ranged between 4.8 (Somalia) to 7.2 well-child visits (Vietnam) (Figure 3-4). In contrast to immunizations, the unadjusted prevalence of

Chapter 3 Study 2 inadequate well-child visits among 10 of 14 immigrant MCOB groups was higher than among children of non-immigrants (17.3%). Somalia had the highest prevalence of inadequate visits

(45.5%), and Vietnam had the lowest (10.8%) (Figure 3-5).

Figure 3-4 includes unadjusted and adjusted mean differences in the number of visits and Figure

3-5 shows the PRs and aPRs of inadequate well-child visits for each MCOB compared to

Canada. After covariate adjustment, infants of mothers born in Jamaica, USA, Pakistan, and

Somalia had significantly fewer visits than infants of non-immigrants. As for inadequate visits, infants of mothers born in 7 of 14 immigrant MCOB, including Sri Lanka, India, Jamaica, USA,

China, Pakistan, and Somalia were 3-84% more likely than children of non-immigrants to have had less than five visits.

Infants of Vietnam and Philippines born mothers were at reduced risk of inadequate well-child care while infants of Polish, Iraqi, Guyanese, Afghanistan, and Iranian mothers did not differ from non-immigrants.

Younger maternal age, increased birth order, lower income quintile, and rurality were associated with decreases in both the number of immunizations and well-child visits and increases in the prevalence of under-immunization and inadequate well-child visits (results not shown).

Chapter 3 Study 2

3.4.4 Sensitivity Analyses Sensitivity analyses revealed that the included and excluded cohort differed on a few factors, including a higher proportion of younger mothers, higher order births, low-income quintile and rural families in the excluded cohort (Appendix 3-2). The method to estimate RSEs did not alter results (Appendix 3-8). Stratifying the cohort by different immunization eras (2002-2004 and

2005-2013) and the era with complete CHC data (2008-2013) generated moderately different effects for some groups (Appendix 3-9). In contrast to the primary results, in 2002-2004 birth years, under-immunization effects for Somalia, USA, and Poland attenuated to the null effect.

Effects for Jamaica on well-child visits attenuated to the null using complete CHC data only.

Results using uncorrected immunization data were equivalent to the primary results using corrected immunization data (Appendix 3-10). Finally, we checked if inadequate immunization using recommendations by 12 months of age would produce the same results as 24 months

(Appendix 3-11). The effect of Somalia and the USA on under-immunization attenuated to the null.

3.5 Discussion

3.5.1 Summary of findings This study linked population-level data sources to examine differences in under-immunization and inadequate well-child visits among Ontario-born infants of various MCOB. Infants of Polish,

American, and Somali-born mothers were more likely to be under-immunized than those of

Canadian mothers. All other immigrant groups were less likely to be under-immunized than non- immigrants. Children of half the foreign MCOB were more likely to have inadequate well-child visits compared to Canada. While previous studies report different routine primary care for

Chapter 3 Study 2 children of immigrants compared to non-immigrants, little research has focused on more than one measure of infant preventive care delineated by several maternal countries of birth.

3.5.2 Interpretation Our results contrast with a recently published systematic review where most included studies demonstrated lower vaccination rates among immigrant children. However, in line with our findings, immigrant children had less access to and utilization of primary care compared to non- immigrants [12].

Vaccine hesitancy due to distrust in medical, public health, and government systems [28] is common in Canada [29] as well as among higher income, higher parity, and white households in other Western countries [30, 31]. Similar mechanisms may exist among parents born in Poland,

America and Somalia. For example, qualitative research from Sweden indicates vaccine hesitancy among Somali families due to the belief that vaccines can cause side effects such as autism and result in postponing or refraining from vaccination [32]. In fact, there have been recent outbreaks in the Somali-American population in Minnesota [33]. Qualitative research is needed to best understand mechanisms behind vaccine hesitancy to ensure infants receive necessary immunizations in each group [34, 35].

We found a higher risk of inadequate well-child visits among several immigrant groups including those from Somalia, Pakistan, China, India, Jamaica, and Sri Lanka. Preventive primary care systems are still developing in some source countries, and remain fragmented with low penetration [36, 37], whereas immunization coverage is increasing due to innovation in vaccine delivery [38]. Quantitative and qualitative research has demonstrated that immigrant parents may have lower awareness of available preventive services [4, 39] and may face greater

Chapter 3 Study 2 difficulties obtaining regular appointments for their children due to limited physician supply

[22], transportation, and child care difficulties [6, 40].

Physicians may be administering more vaccinations at fewer well-child visits if recommended visits are missed [41]. More research is needed to understand parental decision-making for anticipatory well-child care among various immigrant groups as well as in the general population.

3.5.3 Strengths To our knowledge, this was the first retrospective population-based cohort study to examine multiple routine primary care outcomes among children of immigrants. Due to the substantial diversity of the Ontario population, we were able to delineate our outcomes by many MCOB

[42]. The use of population-based health care data eliminates many external validity issues seen in smaller studies; however, immigrant groups in different contexts may display different patterns of primary care. The use of an official government immigration registry improves internal validity of MCOB.

3.5.4 Limitations Here we address several study limitations. First, immigrants who landed prior to 1985 were misclassified as non-immigrants, however, given their long duration of residence outcomes may be closer to Canadian norms. Second, we are unable to assess paternal country of birth, potentially biasing results towards the null effect. Third, all the census-derived variables are measured infrequently and may change over time, resulting in residual confounding [44]. Fourth, excluded children had a more adverse socioeconomic profile, which might indicate that we may not have captured those most at risk. Fifth, we were unable to adjust for several confounders

(e.g., duration of residence since landing, refugee status) where there was no variation in the

Chapter 3 Study 2 non-immigrant group. The Somali group had the largest proportion of refugees, which may help to explain the adverse outcome rates in that group. Additionally, we adjusted for neighborhood income quintile in the analysis as socioeconomic position is related to birthplace as well as healthcare behaviours. However, we may have over adjusted for this factor as it may also be on the causal pathway between birthplace and the study outcomes, as seen by the attenuation of differences with the Somali group. Comparison with crude outcomes in other groups revealed minimal meaningful differences however. Finally, findings may not be generalizable to children born outside of Ontario.

Two other considerations relate to OHIP billing completeness. Nurse practitioners have an expanding scope of practice in Ontario, for which billing data is mostly unavailable [45]. As well, we found lower immunization than earlier reports that ascertained coverage with surveys and immunization records [46]. In the Ontario portion of the Childhood National Immunization

Coverage Survey, antigen-specific coverage at two years of age ranged from 75.0% (Hib) to

93.4% (Polio) [47]. However, the goal of our study was not to estimate antigen-specific coverage but to approximate a measure of routine health care utilization, using overall vaccine dose counting. One study using physician billing for dose-counting found a similar prevalence (42%) of under-immunization to ours [48].

3.5.5 Implications Research is scarce exploring the receipt of routine care for the young children of immigrants of varied MCOB. In a universal health care setting like Canada, the current research suggests

MCOB differences in the number of immunizations and well-child visits. Due to the increasing diversity of immigrant groups, exploring child health care patterns across varied MCOB is of growing importance. Research is needed to unpack mechanisms behind infant routine care

Chapter 3 Study 2

MCOB differences, and to develop and implement targeted interventions to address inadequate care.

Chapter 3 Study 2

Table 3-1. Descriptive demographic characteristics of mother-infant sets among the 15 maternal countries of birth with the greatest share of live deliveries in Ontario, April 2002- March 2013

Maternal Country of Birth Canada India China Pakistan Philippines Sri Lanka Jamaica Poland Vietnam Afghanistan Guyana Iraq Iran USA Somalia N (%) 880,004 43,729 31,903 24,677 21,835 17,444 10,530 7971 8083 6379 5881 5449 5408 5282 5138 N total=1,079,713 (81.5) (4.1) (3.0) (2.29) (2.0) (1.62) (0.98) (0.74) (0.75) (0.6) (0.5) (0.5) (0.5) (0.5) (0.5) Maternal Age % <19 4.2 0.3 0.2 0.7 1.2 0.3 5.2 1.0 0.5 1.5 2.3 1.3 0.4 1.5 1.3 20-34 76.1 87.7 68.4 84.2 62.3 82.5 73.3 72.2 76.0 82.3 77.0 75.0 64.0 71.8 69.9 35+ 19.7 12.0 31.3 15.0 17.2 21.4 21.4 19.8 23.4 16.0 20.6 23.6 35.5 26.5 28.4 Female Gender % 49.0 47.7 48.5 49.1 48.8 50.7 49.3 48.3 48.6 49.7 48.3 49.2 49.1 48.3 49.8 Birth Order % 1st 48.6 58.8 68.3 51.7 60.5 49.2 47.1 54.4 50.6 44.7 47.8 48.6 69.9 52.9 27.1 2nd 35.8 32.8 27.6 31.8 30.2 37.4 32.7 35.2 37.3 32.2 36.9 31.4 27.6 30.6 22.6 3rd 11.5 7.2 3.7 12.7 7.8 11.8 13.7 8.5 9.8 15.6 11.7 14.5 2.2 10.0 19.0 4th 4.1 1.3 0.4 3.8 1.6 1.7 6.5 1.9 2.3 7.4 3.6 5.6 0.3 6.5 31.4 Maternal Time Since

Landing in Canada % <5 years - 63.8 65.2 66.9 51.7 54.3 18.4 19.5 39.3 47.2 25.4 58.1 51.5 63.8 33.5 5-9 years - 24.7 27.2 22.7 22.9 23.2 23.3 14.9 26.2 28.0 26.1 23.8 25.3 17.3 30.0 9-14 years - 7.9 5.9 7.8 15.2 13.6 28.7 25.2 18.6 16.0 23.5 12.2 13.0 8.6 25.7 15+ years - 3.7 1.7 3.0 10.2 8.8 29.7 40.4 15.9 8.8 25.0 5.9 10.2 10.4 10.8 Neighborhood Income

Quintile % 1: Lowest 17.9 29.7 27.9 40.1 33.3 36.2 43.3 16.3 33.0 50.3 35.1 43.6 19.0 14.0 69.9 2 18.8 27.1 28.4 22.3 24.7 28.8 23.6 19.0 24.8 16.7 25.1 21.0 15.2 17.2 15.6 3 20.7 23.5 18.6 17.7 19.5 20.3 19.0 21.3 21.7 13.2 23.0 16.4 21.4 19.5 7.3 4 22.6 13.7 0.4 14.5 14.4 10.8 9.8 26.1 14.0 13.5 12.1 12.8 26.6 22.4 5.0 5: Highest 19.3 5.8 9.3 5.3 8.0 3.7 4.3 17.2 6.4 6.2 4.6 5.9 17.4 26.7 2.0 Lives in Rural Area % 13.3 0.3 0.3 0.3 0.8 0.1 0.1 1.7 0.3 0.2 0.2 0.2 0.2 9.5 0.0 Mother is a Refugee % - 1.2 6.4 6.0 0.1 24.7 0.4 23.2 9.4 51.8 2.2 37.9 19.4 2.4 63.9

Chapter 3 Study 2

Figure 3-1 Flow Chart of Cohort Formation

At least one well-child visit Excluded due to no well- and one immunization child visits or no N=1,079,713 immunizations Final Cohort N=82,050

Chapter 3 Study 2

Figure 3-2 Unadjusted ■ and adjusted ● mean difference in the number of vaccine doses among Ontario-born children by age 2 years across maternal countries of birth, compared to Canadian-born mothers. Adjustment was for maternal age, birth year, neighborhood income quintile, birth order, child gender, and rurality.

Chapter 3 Study 2 Figure 3-3 Unadjusted ■ and adjusted ● prevalence ratios of under-immunization among Ontario-born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers. Adjustment was for maternal age, birth year, neighborhood income quintile, birth order, child gender, and rurality.

(43.10)

Chapter 3 Study 2 Figure 3-4 Unadjusted ■ and adjusted ● mean difference in the number of well-child visits among Ontario- born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers. Adjustment was for maternal age, birth year, neighborhood income quintile, birth order, child gender, and rurality.

Chapter 3 Study 2 Figure 3-5 Unadjusted ■ and adjusted ● prevalence ratios of inadequate well-child visits among Ontario- born children at 2 years of age across maternal countries of birth, compared to Canadian-born mothers. Adjustment was for maternal age, birth year, neighborhood income quintile, birth order, child gender, and rurality.

Chapter 3 Study 2 3.6 Appendices

Appendix 3-1 Distribution of exclusion criteria across maternal countries of birth

Maternal country of Birth Overall Canada India China Pakistan Philip. Sri Lanka Jamaica Poland Vietnam Afghanistan Guyana USA Iraq Iran Somalia Number of births 1,410,572 1,157,865 55,215 39,339 31,716 27,349 22825 13 620 9811 9864 7944 7827 6945 6778 6653 6830 N (%) Number 330,859 277,861 11,486 7427 7039 5514 5381 3090 1840 1781 1565 1946 1163 1329 1245 1692 excluded N(%) (23.5) (24) (20.8) (18.9) (22.2) (20.2) (23.6) (22.7) (18.8) (18.1) (19.7) (24.9) (24.0) (19.6) (18.7) (24.8) Birth 158,758 11.3 11.9 7.7 11.3 13.2 11.9 14.5 9.2 9.6 8.4 17.3 11.4 10.5 9.6 9.4 exclusion (11.3) Complex 76,185 Chronic (5.4) 5.3 6.4 4.1 6.4 6.5 6.0 8.3 4.1 4.6 6.0 8.4 4.8 4.7 5.1 5.7 Conditions % Community 34,774 Health Centre (2.5) 2.6 1.5 1.5 2.3 1.1 1.3 2.6 1.8 2.4 5.2 1.8 2.0 2.6 1.5 6.5 Use % Death before 5556 0.4 0.4 0.2 0.5 0.4 0.4 0.9 0.3 0.4 0.6 0.6 0.4 0.4 0.3 0.6 24 months % (0.4) Incomplete 40,175 OHIP (2.9) 2.9 2.8 2.4 3.6 2.3 2.4 2.9 2.4 2.2 3.2 2.5 2.8 2.8 2.5 4.0 Eligibility % Incomplete 8645 0.7 0.4 0.6 0.4 0.1 0.2 0.1 0.2 0.1 0.2 0.1 0.5 0.4 0.6 0.3 covariate data % (0.61) No documented 134,732 well-child visits or (9.6) 10.1 5.8 7.9 7.4 4.5 9.9 6.6 7.5 5.9 7.5 5.9 10.1 6.8 5.9 12.9 immunizations %

Chapter 3 Study 2

Appendix 3-2 Comparison of mother-infant sets included in the cohort to those excluded due to no documented immunizations or well-child care

Included Cohort Excluded cohort Standardized Characteristic n=1,079,713 (92.9%) n=82,050 (7.1%) Difference* Maternal Age % <19 35703 (3.3) 5261 (6.4) -0.14 20-34 823438 (76.3) 62976 (76.8) -0.01 35+ 219759 (20.4) 13690 (16.7) 0.09 Female Gender % 528727 (49.0) 40220 (49.0) 0 Birth Order % 1st 541491 (50.2) 38732 (47.2) 0.06 2nd 380916 (35.3) 27047 (33.0) 0.05 3rd 118378 (11.0) 10299 (12.6) -0.05 4th or more 38928 (3.6) 5972 (7.3) -0.16 Maternal Duration Since Landing in Canada % N/A 879997 (81.5) 71519 (87.2) -0.16 <5 years 108322 (10.0) 6161 (7.5) 0.09 5-9 years 48450 (4.5) 2352 (2.9) 0.09 9-14 years 25079 (2.3) 1150 (1.4) 0.07 15+ years 17865 (1.7) 868 (1.1) 0.05 Neighborhood Income Quintile % 1: Lowest 220063 (20.4) 21851 (26.6) -0.15 2 215115 (19.9) 17020 (20.7) -0.02 3 224599 (20.8) 15566 (19.0) 0.05 4 212155 (21.5) 15080 (18.4) 0.08 5: Highest 187781 (17.4) 12533 (15.3) 0.06 Lives in Rural Area % 103907 (9.6) 21602 (26.3) -0.45 Mother is a Refugee % 20995 (1.9) 1261 (1.5) 0.03 *Standardized difference = difference in proportions between included and excluded infants on the basis of documented visits and immunizations, divided by standard error; imbalance defined as an absolute value greater than 0.10

Chapter 3 Study 2 Appendix 3-3 Measurement of Well-Child Visits

Variable Description Well-child visits • Variable counting the frequency of well-child well child visits (max 1 per day (SERVDATE)) occurring within the first 2 years of life (6 weeks to 24.5 months): 1. EITHER Feecode A007 + dx code 916, 917, 896, or 999 2. OR Feecode A262 + dxcode 916, 917, 896, or 999 3. OR Feecode A001 + dxcode 916, 917, 896, or 999 4. OR Feecode A003 + dxcode 916, 917, 896, or 999 5. OR Feecode A261 + dxcode 916, 917, 896, or 999 6. OR Feecode A903 + dxcode 916, 917, 896, or 999 7. OR Feecode K013 + dxcode 916, 917, 896, or 999 8. OR Feecode A004 + dxcode 916, 917, 896, or 999 9. OR Feecode K119 + dxcode 916, 917, 896, or 999 10. OR Feecode A008 + dxcode 916, 917, 896, or 999 11. OR Feecode A007+ Feecode immunization on same SERVDATE 12. OR Feecode A002 13. OR Feecode A268 14. OR Feecode K017 15. OR Feecode K267 If eligible infant has no relevant OHIP claims, then the child is excluded Description Fee codes Fee code Meaning DXcodes dxmeaning 1. EITHER Fee code A007 AND A007 Intermediate Assessment or Well-child Care 916 Well-child dx code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 2. OR Fee code A262 AND dx A262 level 2 pediatric assessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 3. OR Fee code A001 AND dx A001 Minor assessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 4. OR Fee code A003 AND dx A003 General assessment Family practice and practice in 916 Well-child code 916, or 917, or 896, or general 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 5. OR Fee code A261 AND dx A261 level 1 pediatric assessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 6. OR Fee code A903 AND dx A903 pre-dental/pre-operative general assessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 7. OR Fee code K013 AND dx K013 family practice and practice in general mental health 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 8. OR Fee code A004 AND dx A004 general reassessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult 88

Chapter 3 Study 2 ‘999.' 896 Immunization 999 W/o diag 9. OR Fee code K119 AND dx K119 pediatric developmental assessment incentive 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 10. OR Fee code A008 AND dx A008 mini assessment 916 Well-child code 916, or 917, or 896, or 917 Well-adult ‘999.' 896 Immunization 999 W/o diag 11. OR Fee code A007 AND Fee A007 AND (G538, Intermediate Assessment or Well-child Care + Any N/A - code immunization on same day G539, G840, G841, immunization fee code G844, G845, G846, G847, or G848) 12. OR Fee code A002 A002 Enhanced 18 month well-child visit (family doctor) - - 13. OR Fee code A268 A268 Enhanced 18 month well-child visit (pediatrician) - - 14. OR Fee code K017 K017 Periodic Health visit child - - 15. OR Fee code K267 K267 Periodic health visits 2-11 years of age - - Inadequate Well- The infant has attended less than the minimum targeted 5 well-child visits. This helps to identify children who are not receiving the recommended care (5/5 visits). Yes child visits or No

Appendix 3-4 Measurement of Immunizations

Chapter 3 Study 2

Variable Description Fee Fee code Meaning DXcodes DXmeaning codes Number of • Variable counting the total number of immunizations received by an infant between 6 G538 Other immunizing N/A immunizations weeks to 24.5 months of age agents not listed delivered by a pediatrician or general practitioner below • If the fee code is generic (G538, or G539) can be counted more than once per date of G539 Other immunizing - service (SERVDATE) and each occurrence is multiplied by the number of units delivered agents not listed (NUMSERV). If the fee code is antigen-specific, it can occur at a maximum of one specific below fee code/day. See Decision Tree Below in Appendix 2C. G840 DTaP-IPV - • If an eligible infant has no relevant ohip claims, then the child is excluded; if the number is G841 DTAP-IPV-HiB - an outlier above the 99th percentile then the number of immunizations=. G844 Men-C - G845 MMR - G846 DTAP-IPV-HiB - G847 Tdap - G848 VAR - A007 Assessment or 896 Immunization well-child care Under- The infant has received less than the number of publicly funded immunization in the infant's birth year, or the following year if the immunization schedule increased immunization its provision. Yes or No.

Chapter 3 Study 2 Appendix 3-5 Decision tree for counting immunization billing codes

Claims on different Count by adding up SERVDATE NUMSERV/claim/SERVDATE

Generic codes G538 or G539 Count by adding up

Different FEECODE NUMSERV/claim

Multiple claims on the same SERVDATE Count by adding up

Same FEECODE NUMSERV/claim

Claims on different Count as one per individual SERVDATE claim, not NUMSERV

Antigen-specific codes G840, G841, G844, G845, G846, G847, or G848 Count unique antigen-specific Different antigen-specific FEECODEs/day (not NUMSERV) FEECODE Multiple claims on the same SERVDATE

Count as one/SERVDATE (max one Same antigen-specific FEECODE unique antigen-specific FEECODE/SERVDATE)

Chapter 3 Study 2

Appendix 3-6 Handling outliers in the number of immunizations

• The handling of outliers is of significance because there is a limit to the number of immunizations a child can receive realistically. • Table 3 below articulates the decision rules. • Counts > 16 immunizations (>99th pctl.) were assumed to be erroneous and excluded. • In general, counts within a reasonable range of 12-16 immunizations, were corrected to the maximum # of publicly funded in the infants’ birth year plus 2. • These corrections were intended to preserve sample size, within a clinically realistic range • Two assumptions made with the corrections: o 1. The # of publicly funded immunizations plus two is 1. clinically realistic o 2. Or is due to minor data entry error o ≥ Two extra immunizations above those publicly funded after 6 weeks, are less clinically realistic and more likely due to a more substantial data entry error. • The exception to the corrections rules for infants born in 2002-2003: o Infants born in these years may ‘catch-up' in the second year of life when the immunization schedule increased • These decision rules result in: o N excluded infants= 12,748 (0.76%) o N corrected infants= 80,228; n corrected to 12=32,164 and n corrected to 14=48,064 Decision rules for immunization outliers and correction Birth year # of publicly funded A. # immunizations B. # immunizations range immunizations counted C. corrected 2002-2013 Ranges from 5 to 12 >16 Missing 2002-2013 Ranges from 5 to 12 0 to 12 Not corrected- Same as A. 2002-2003 5 12 to 14 12 2002-2003 5 15 to 16 12 2004-2009 12 12 to 14 Not corrected- Same as A. 2004-2009 12 15 to 16 14 (12 +2 extra) 2010 11 11 to 13 Not corrected- Same as A. 2010 11 14 to 15 13 (11+2 extra) 2011-2013 10 12 to 14 12 (10 +2 extra) 2011-2013 10 15 to 16 12 (10+2 extra)

Chapter 3 Study 2

Appendix 3-7 Distribution of the number well-child visits and immunizations by infant fiscal year of birth, April 2002- March 2014

Birth Year # Publicly Mean # Mode 0 10 25 50 75 90 99 Low High Under Funded Immunizations Immunized % Vaccine (SD) Doses 2002 5 6.5 (3.2) 5 1 3 5 5 8 11 13 1 108 23.3 2003 5 7.4 (3.5) 5 1 4 5 7 10 12 14 1 97 18.6 2004 5 9.3 (3.6) 12 1 3 5 10 12 13 18 1 100 9.5 2005 12 10.0 (3.4) 12 1 5 8 11 12 13 18 1 109 57.4 2006 12 10.2 (3.3) 12 1 6 8 11 12 14 18 1 73 54.7 2007 12 10.2 (3.4) 12 1 6 8 11 12 14 18 1 104 54.4 2008 12 10.2 (3.4) 12 1 6 8 11 12 14 18 1 110 53.0 2009 12 10.2 (3.4) 12 1 5 8 11 12 14 18 1 75 52.3 2010 12 9.6 (3.1) 11 1 5 8 10 11 13 17 1 64 37.7 2011 10 9.2 (2.9) 10 1 5 8 10 11 12 16 1 96 42.8 2012 10 9.3 (2.8) 10 1 5 8 10 11 12 16 1 91 39.1 2013 10 9.5 (2.8) 10 1 6 8 10 11 12 16 1 30 35.5 Mean # Well- % Inadequate Visits Child Visits (SD) 2002 6.3 (2.3) 6 1 3 5 6 8 9 13 1 31 18.9 2003 6.5 (2.4) 6 1 4 5 6 8 10 11 1 34 17.4 2004 6.9 (2.4) 7 1 4 5 7 8 10 13 1 23 14.1 2005 6.7 (2.3) 6 1 4 5 7 8 9 11 1 36 14.3 2006 6.6 (2.2) 7 1 3 5 7 8 9 13 1 25 15.5 2007 6.4 (2.2) 7 1 4 5 6 8 9 12 1 23 18.1 2008 6.4 (2.3) 7 1 3 5 6 8 9 12 1 34 18.1 2009 6.4 (2.2) 7 1 3 5 6 8 9 12 1 31 17.9 2010 6.2 (2.2) 6 1 3 5 6 7 9 12 1 28 19.0 2011 6.1 (2.2) 6 1 3 5 6 7 9 12 1 32 20.7 2012 6.1 (2.2) 7 1 3 5 6 7 9 12 1 22 20.3 2013 6.2 (2.6) 7 1 2 5 6 7 9 12 1 21 19.8

Chapter 3 Study 2

Appendix 3-8 Differences in under-immunization by maternal country of birth, 2002-2013 using three methods to estimate robust standard errors

Under-immunization Inadequate Well-child visits Test Case 2- Base Case- Test Case 1- Base Case Test Case 1- Test Case 2- Infant ID (Maternal Maternal ID Infant ID Maternal ID Infant ID Infant ID (Maternal ID) ID) Maternal Country PR 95% CI PR 95% CI PR 95% CI PR 95% CI PR 95% CI PR 95% CI of Birth Canada (referent) 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 India 0.94 (0.93, 0.95) 0.94 (0.93, 0.95) 0.94 (0.93, 0.95) 1.03 (1.01, 1.05) 1.04 (1.02, 1.05) 1.04 (1.02, 1.05) China 0.78 (0.77, 0.80) 0.78 (0.77, 0.79) 0.78 (0.77, 0.79) 1.19 (1.17, 1.22) 1.19 (1.17, 1.22) 1.19 (1.17, 1.22) Pakistan 1.03 (1.01, 1.04) 1.02 (1.01, 1.03) 1.02 (1.01, 1.03) 1.27 (1.24, 1.30) 1.25 (1.22, 1.27) 1.25 (1.22, 1.27) Philippines 0.75 (0.73, 0.76) 0.75 (0.73, 0.76) 0.75 (0.73, 0.76) 0.80 (0.77, 0.82) 0.80 (0.78, 0.83) 0.80 (0.78, 0.83) Sri Lanka 0.79 (0.77, 0.81) 0.79 (0.78, 0.81) 0.79 (0.78, 0.81) 1.14 (1.11, 1.17) 1.14 (1.11, 1.17) 1.14 (1.11, 1.17) Jamaica 0.96 (0.94, 0.98) 0.96 (0.94, 0.98) 0.96 (0.94, 0.98) 1.05 (1.01, 1.09) 1.06 (1.02, 1.09) 1.06 (1.02, 1.09) Poland 1.06 (1.04, 1.09) 1.06 (1.04, 1.09) 1.06 (1.04, 1.09 0.99 (0.94, 1.04) 0.99 (0.95, 1.03) 0.99 (0.95, 1.03) Vietnam 0.66 (0.64, 0.68) 0.66 (0.64, 0.68) 0.66 (0.64, 0.68) 0.61 (0.57, 0.64) 0.61 (0.57, 0.64) 0.61 (0.57, 0.64) Afghanistan 0.87 (0.84, 0.89) 0.86 (0.84, 0.89) 0.86 (0.84, 0.89) 0.88 (0.83, 0.93) 0.86 (0.82, 0.91) 0.86 (0.82, 0.91) Guyana 0.88 (0.85, 0.91) 0.88 (0.86, 0.91) 0.88 (0.86, 0.91) 0.91 (0.86, 0.96) 0.92 (0.87, 0.97) 0.92 (0.87, 0.97) USA 1.06 (1.04, 1.09) 1.06 (1.04, 1.09) 1.06 (1.04, 1.09) 1.17 (1.12, 1.23) 1.17 (1.12, 1.22) 1.17 (1.12, 1.22) Iraq 0.87 (0.85, 0.90) 0.87 (0.84, 0.90) 0.87 (0.84, 0.90) 1.02 (0.97, 1.08) 0.99 (0.95, 1.04) 0.99 (0.95, 1.04) Iran 0.76 (0.73, 0.79) 0.76 (0.73, 0.78) 0.76 (0.73, 0.78) 1.01 (0.95, 1.07) 1.01 (0.95, 1.06) 1.01 (0.95, 1.06) Somalia 1.09 (1.06, 1.12) 1.07 (1.04, 1.10) 1.07 (1.04, 1.10) 1.73 (1.12, 1.23) 1.67 (1.62, 1.72) 1.67 (1.62, 1.72)

Chapter 3 Study 2

Appendix 3-9 Sensitivity Analyses Differences in Routine Care Outcomes by Maternal Country of Birth, Restricted Birth Years

Inadequate Under immunization Under immunization well-child visits 2002-2004 2005-2013 2008-2013 Birth Years Only Birth Years Only Birth Years Only Maternal Country of Birth PR 95%CI PR 95%CI PR 95%CI Canada (referent) 1.00 1.00 1.00 India 0.96 (0.92-1.04) 0.92 (0.91, 0.93) 0.94 (0.92, 0.97)† China 0.97 (0.92-1.04) 0.74 (0.73, 0.75) 1.17 (1.16, 1.22) Pakistan 1.19 (1.12-1.26)† 0.97 (0.96, 0.99) 1.11 (1.08, 1.14) Philippines 0.65 (0.60-0.71) 0.75 (0.74, 0.77) 0.77 (0.74, 0.80) Sri Lanka 0.56 (0.51-0.62) 0.80 (0.79, 0.82) 1.30 (1.25, 1.34) Jamaica 1.05 (0.97-1.14) 0.93 (0.91, 0.95) 1.00 (0.96, 1.05)† Poland 0.97 (0.86-1.09)† 1.05 (1.02, 1.07) 1.05 (0.99, 1.11) Vietnam 0.44 (0.38-0.52) 0.67 (0.65, 0.69) 0.59 (0.54, 0.64) Afghanistan 0.91 (0.79-1.03) 0.84 (0.82, 0.87) 0.80 (0.75, 0.86) † Guyana 0.92 (0.82-1.03) 0.86 (0.84, 0.89) 0.87 (0.81, 0.94) † USA 1.03 (0.89-1.17)† 1.05 (1.02, 1.08) 1.18 (1.11, 126) Iraq 0.70 (0.59-0.84) 0.87 (0.85, 0.90) 1.00 (0.94, 1.07) Iran 0.70 (0.59-0.84) 0.74 (0.71, 0.77) 0.94 (0.87, 1.02) Somalia 0.96 (0.87-1.07)† 1.08 (1.05, 1.10) 1.81 (1.11, 1.26) † Indicates a qualitative change from the main analyses, either to the null effect or the opposite direction

Chapter 3 Study 2

Appendix 3-10 Under-immunization using uncorrected data from all eligible births 2002-2013

Prevalence of under-immunization on

uncorrected data

Maternal Country of Birth PR 95% CI Canada (referent) 1.00 - India 0.94† (0.93, 0.95) China 0.76† (0.75, 0.77) Pakistan 1.03‡ (1.01, 1.04)‡ Philippines 0.74† (0.73, 0.75) Sri Lanka 0.79† (0.77, 0.80) Jamaica 0.96† (0.94, 0.98) Poland 1.07‡ (1.04, 1.09)‡ Vietnam 0.65† (0.63, 0.67) Afghanistan 0.87† (0.84, 0.89) Guyana 0.88† (0.85, 0.91) USA 1.06‡ (1.04, 1.09)‡ Iraq 0.87† (0.85, 0.90) Iran 0.75† (0.72, 0.78) Somalia 1.09‡ (1.06, 1.12)‡ Results of analyses on uncorrected data are equivalent to the results on uncorrected data † Indicates significantly more care compared to the reference group ‡ Indicates significantly less care compared to the reference group

Chapter 3 Study 2

Appendix 3-11 Differences in the number of immunizations and prevalence of under immunization among infants at 12 months old by maternal country of birth

Prevalence Ratio of Under Immunization 12 months Maternal Country of Birth % PR (95% CI) aPR (95% CI) Canada (referent) 26.2 1 1 Vietnam 19.4 0.74 (0.71, 0.78)† 0.75 (0.71, 0.79)† China 18.4 0.70 (0.68, 0.72)† 0.73 (0.71, 0.75)† Philippines 19.2 0.73 (0.71, 0.76)† 0.75 (0.73, 0.77)† Iran 19.3 0.74 (0.69, 0.78)† 0.79 (0.75, 0.83)† Sri Lanka 26.2 1.00 (0.96, 1.06) 1.00 (0.97, 1.08) Iraq 25.0 0.95 (0.90, 1.01) 0.92 (0.87, 0.97)† India 26.5 1.01 (0.99, 1.03) 1.02 (1.00, 1.03) Guyana 22.9 0.88 (0.83, 0.92)† 0.88 (0.83, 0.92)† Afghanistan 25.4 0.97 (0.92, 1.02) 0.91 (0.87, 0.96)† Pakistan 29.4 1.12 (1.10, 1.15)‡ 1.11 (1.08, 1.13)‡ Jamaica 25.5 0.97 (0.94, 1.01) 0.94 (0.91 ,0.98)† USA 26.5 1.01 (0.96, 1.06) 1.03 (0.98, 1.08) Poland 27.6 1.05 (1.01, 1.10)‡ 1.09 (1.05, 1.13)‡ Somalia 29.7 1.13 (1.08, 1.19)‡ 1.02 (0.97, 1.03) At 12 months children born in 2002-2004 should receive 3 immunizations, and those born after should receive 5. PR prevalence ratio; aPR adjusted prevalence ratio; 95%CI 95% Confidence Interval Adjusted for maternal age, birth order, infant gender, neighborhood income quintile, rurality ‡ indicates higher prevalence of under-immunization † indicates lower prevalence of under-immunization

Chapter 3 Study 2

3.7 References 1. Flores, G., M. Abreu, and S.C. Tomany-Korman, Limited English proficiency, primary language at home, and disparities in children's health care: how language barriers are measured matters. Public health reports, 2005. 120(4): p. 418-430. 2. McKeary, M. and B. Newbold, Barriers to care: The challenges for Canadian refugees and their health care providers. Journal of Refugee Studies, 2010. 23(4): p. 523-545. 3. Ferrer, A.M., G. Picot, and W.C. Riddell, New directions in immigration policy: Canada's evolving approach to the selection of economic immigrants. International Migration Review, 2014. 48(3): p. 846-867. 4. Derose, K.P., J.J. Escarce, and N. Lurie, Immigrants and health care: sources of vulnerability. Health affairs, 2007. 26(5): p. 1258-1268. 5. Pottie, K., et al., Language proficiency, gender and self-reported health: an analysis of the first two waves of the longitudinal survey of immigrants to Canada. Canadian Journal of Public Health/Revue Canadienne de Sante'e Publique, 2008: p. 505-510. 6. Asanin, J. and K. Wilson, “I spent nine years looking for a doctor”: exploring access to health care among immigrants in Mississauga, Ontario, Canada. Social science & medicine, 2008. 66(6): p. 1271-1283. 7. Brown, K.F., et al., Factors underlying parental decisions about combination childhood vaccinations including MMR: a systematic review. Vaccine, 2010. 28(26): p. 4235-4248. 8. Borràs, E., et al., Parental knowledge of paediatric vaccination. BMC Public Health, 2009. 9(1): p. 154. 9. Guttmann, A., et al., Immunization coverage among young children of urban immigrant mothers: findings from a universal health care system. Ambulatory Pediatrics, 2008. 8(3): p. 205-209. 10. Borras, E., et al., Vaccination coverage in indigenous and immigrant children under 3 years of age in Catalonia (Spain). Vaccine, 2007. 25(16): p. 3240-3243. 11. Varan, A.K., et al., Vaccination Coverage Disparities Between Foreign-Born and US- Born Children Aged 19–35 Months, United States, 2010–2012. Journal of immigrant and minority health, 2017. 19(4): p. 779-789. 12. Markkula, N., et al., Use of health services among international migrant children–a systematic review. Globalization and health, 2018. 14(1): p. 52. 13. Feudtner, C., et al., Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. BMC pediatrics, 2014. 14(1): p. 199. 14. Ray, J.G., D.A. Henry, and M.L. Urquia, Sex ratios among Canadian liveborn infants of mothers from different countries. Canadian Medical Association Journal, 2012: p. cmaj. 120165. 15. Urquia, M.L., et al., Sex ratios at birth after induced abortion. Canadian Medical Association Journal, 2016. 188(9): p. E181-E190.

Chapter 3 Study 2

16. Urquia, M.L., et al., Variations in male-female infant ratios among births to Canadian- and Indian-born mothers, 1990-2011: a population-based register study. CMAJ open, 2016. 4(2): p. E116. 17. Chiu, M., et al., Describing the linkages of the immigration, refugees and citizenship Canada permanent resident data and vital statistics death registry to Ontario’s administrative health database. BMC medical informatics and decision making, 2016. 16(1): p. 135. 18. Wilson, K., et al., Patterns of emergency room visits, admissions and death following recommended pediatric vaccinations—a population based study of 969,519 vaccination events. Vaccine, 2011. 29(21): p. 3746-3752. 19. Hawken, S., et al., Association between birth order and emergency room visits and acute hospital admissions following pediatric vaccination: a self-controlled study. PloS one, 2013. 8(12): p. e81070. 20. Salmon, D.A., et al., Measuring immunization coverage among preschool children: past, present, and future opportunities. Epidemiologic reviews, 2006. 28(1): p. 27-40. 21. Canadian Immunization Registry Network. National Standards for Immunization Coverage Assessment: Recommendations from the Canadian Immunization Registry Network, in Canadian Communicable Disease Report. 2012. 22. Guttmann, A., et al., Primary care physician supply and children's health care use, access, and outcomes: findings from Canada. Pediatrics, 2010. 125(6): p. 1119-1126. 23. Canadian Society of Pediatrics. Schedule of well-child visits 2016 [cited 2018 May 30 2018]; Available from: https://www.caringforkids.cps.ca/handouts/schedule_of_well_child_visits. 24. Schwartz, K.L., et al., Validation of infant immunization billing codes in administrative data. Human vaccines & immunotherapeutics, 2015. 11(7): p. 1840-1847. 25. Kralj, B., Measuring “rurality” for purposes of health-care planning: an empirical measure for Ontario. Ont Med Rev, 2000. 67(9): p. 33-52. 26. Zou, G., A modified poisson regression approach to prospective studies with binary data. American journal of epidemiology, 2004. 159(7): p. 702-706. 27. Zou, G. and A. Donner, Extension of the modified Poisson regression model to prospective studies with correlated binary data. Statistical methods in medical research, 2013. 22(6): p. 661-670. 28. Salmon, D.A., et al., Factors associated with refusal of childhood vaccines among parents of school-aged children: a case-control study. Archives of pediatrics & adolescent medicine, 2005. 159(5): p. 470-476. 29. Dubé, E., et al., Parental vaccine hesitancy in Quebec (Canada). PLoS currents, 2016. 8. 30. Smith, P.J., S.Y. Chu, and L.E. Barker, Children who have received no vaccines: who are they and where do they live? Pediatrics, 2004. 114(1): p. 187-195. 31. Siddiqui, M., D.A. Salmon, and S.B. Omer, Epidemiology of vaccine hesitancy in the United States. Human vaccines & immunotherapeutics, 2013. 9(12): p. 2643-2648.

Chapter 3 Study 2

32. Sweden, T.P.H.A.o., Barriers and motivating factors to MMRr vaccination in communities with low coverage in Sweden. 2015, Folkhälsomyndigheten: Stockholm. 33. Gahr, P., et al., An outbreak of measles in an undervaccinated community. Pediatrics, 2014. 134(1): p. e220-e228. 34. Dubé, E., et al., The WHO Tailoring Immunization Programmes (TIP) approach: Review of implementation to date. Vaccine, 2017. 35. Jama, A., et al., Tailored communication interventions targeting Somali community in Sweden regarding MMR vaccination. European Journal of Public Health, 2017. 27:Supplement 3 . 36. Bajpai, N. and S. Goyal, Primary health care in India: Coverage and quality issues. 2004. 37. Mills, A., Health care systems in low-and middle-income countries. New England Journal of Medicine, 2014. 370(6): p. 552-557. 38. World Health Organization, Global routine vaccination coverage, 2016–Couverture de la vaccination systématique dans le monde, 2016. Weekly Epidemiological Record= Relevé épidémiologique hebdomadaire, 2017. 92(46): p. 701-707. 39. Carroll, J., et al., Knowledge and beliefs about health promotion and preventive health care among Somali women in the United States. Health care for women international, 2007. 28(4): p. 360-380. 40. Condon, L. and S. McClean, Maintaining pre-school children's health and wellbeing in the UK: a qualitative study of the views of migrant parents. Journal of Public Health, 2016. 39(3): p. 455-463. 41. Guttmann, A., et al., Volume matters: physician practice characteristics and immunization coverage among young children insured through a universal health plan. Pediatrics, 2006. 117(3): p. 595-602. 42. Pavlish, C.L., S. Noor, and J. Brandt, Somali immigrant women and the American health care system: discordant beliefs, divergent expectations, and silent worries. Social Science & Medicine, 2010. 71(2): p. 353-361. 43. Reitz, J.G., M.B. Phan, and R. Banerjee, Gender equity in Canada's newly growing religious minorities. Ethnic and Racial Studies, 2015. 38(5): p. 681-699. 44. Durbin, A., et al., Examining the relationship between neighbourhood deprivation and mental health service use of immigrants in Ontario, Canada: a cross-sectional study. BMJ open, 2015. 5(3): p. e006690. 45. Hutchison, B. and R. Glazier, Ontario’s primary care reforms have transformed the local care landscape, but a plan is needed for ongoing improvement. Health affairs, 2013. 32(4): p. 695-703. 46. Wilson, S.E., et al., Methods used for immunization coverage assessment in Canada, a Canadian Immunization Research Network (CIRN) study. Human vaccines & immunotherapeutics, 2017. 13(8): p. 1928-1936.

100

Chapter 3 Study 2

47. Public Health Agency of Canada., Vaccine coverage in Canadian children: Results from the 2013 childhood National Immunization Coverage Survey (cNICS). 2016. 48. Dummer, T., et al., Immunization completeness of children under two years of age in Nova Scotia, Canada. Can J Public Health, 2012. 103(5): p. e363-7.

101

Chapter 4 Study 3

Chapter 4, Study 3: Comparison of receipt of routine preventive care among infant daughters and sons of immigrant mothers to Ontario, Canada

102

Chapter 4 Study 3

4.1 Abstract Background

Son preference has been demonstrated among immigrant groups via sex-selective pregnancy terminations, specifically among Punjabi-Canadians. Son preference may similarly adversely affect daughters’ receipt of health care compared to their brothers. The objective of the study was to explore gender disparities in infant routine preventive care across maternal countries of birth

(MCOB), and by mother tongue among infants of Indian-born mothers.

Methods

This a retrospective population-based cohort using administrative databases including all healthy, term singletons with at least one opposite gender sibling born Ontario between April

2002 and March 2014. Using linked administrative databases, we identified 350,366 infants of

154,259 mothers born in the 15 countries most commonly delivering in Ontario. Fixed effects conditional logistic regression generated adjusted odds ratios for a daughter being under- immunized and having an inadequate number of well-child visits compared to her brother, stratified by MCOB. Modifying effects by maternal mother tongue were assessed among families with Indian-born mothers.

Results

Girls whose mothers were born in India had 1.19 times (95% CI: 1.07, 1.33) the adjusted odds of inadequate number of well-child visits versus their brothers. This effect was limited to the

Punjabi mother tongue subgroup (aOR:1.26, 95%CI: 1.08, 1.47).

Conclusions

103

Chapter 4 Study 3

Gender equity in routine preventive health care is mostly achieved among children of immigrants. However, infant daughters of Indian-born mothers whose mother tongue is Punjabi, appear to be at a disadvantage for well-child visits compared to their brothers. This suggests son preference may persist beyond the family planning stage in the Indo-Canadian community.

104

Chapter 4 Study 3

4.2 Introduction In early childhood, parents are the gatekeepers to their child’s health care use [1]. Barriers facing immigrants in concert with health beliefs and family dynamics may influence parental healthcare decision-making around immunizations and well-child check-ups [1-3]. Evidence shows that beliefs about gender norms, roles and relations adversely affect the health and wellbeing of girls in top source countries of immigrants to Canada including India, Pakistan, and China, among others [4-8]. In select countries, differential health investments exist towards sons and daughters, to the disadvantage of girls with respect to breastfeeding, immunization, and seeking health care for illness [9-12], which some researchers have termed ‘health-care neglect’ [8, 13]. Studies have identified son-preference among immigrant communities in Canada and other immigrant receiving-countries manifesting through sex-selective pregnancy termination [5, 14-19]. In the

Indo-Canadian community, it is documented at higher birth orders particularly among mothers whose first language is Punjabi [18].

It is unknown if son preference may affect the routine preventive care of young girls and boys across different immigrant groups. One study of immigrants to the United Kingdom identified insignificant gender differences in immunization rates. However, that study had a limited sample size and did not examine gender bias within families [20]. Daughters in some immigrant groups may experience ‘double jeopardy’ concerning health care in early life due to their gender and parental country of birth, and such disparities must be quantified [21].

Ontario, Canada provides an ideal setting in which to conduct health research on the children of immigrants. Ontario has one of the most diverse immigrant populations in the world [22], and children are covered for universal health care in Canada from birth including routine anticipatory

105

Chapter 4 Study 3 care (i.e., check-ups and the recommended series of vaccinations) by a publicly-funded health insurance system, without direct cost to the parent.

The primary objective of the study was to identify any existing gender disparities in routine preventive care within families across various maternal countries of birth. Since evidence has demonstrated sex-selective pregnancy termination within specific linguistic subgroups in the

Indo-Canadian community, the secondary objective was to investigate if gender disparities among this subgroup were dependent on mother tongue.

4.3 Methods

4.3.1 Data sources The data for this study comes from several linked population-based administrative databases at

ICES. A unique coded identity number facilitates record linkage between the databases. The

Registered Persons Database (RPDB) is the provincial health care registry. It contains information on birth date, sex, and postal code which was linked to Canadian Census data to obtain neighborhood information at the level of a dissemination area.

Hospitalization-related deliveries between April 1st, 2002 and March 31st, 2014 were identified from the Discharge Abstract Database (DAD) of the Canadian Institute for Health Information

(capturing 98% of births). Well-child check-ups and vaccinations with family physicians and pediatricians were captured using The Ontario Health Insurance Plan (OHIP) claims database, which contains information on physicians’ billings, such as fee codes for visits, diagnostic codes, and date of service. We sourced the Ontario portion of the federally maintained Immigration,

Refugees, and Citizenship Canada (IRCC) Permanent Resident Database for information on maternal birthplace, immigrant class, and landing date to Canada. Many earlier studies have linked the IRCC databases with the other databases used in this study [14, 16, 23]. Overall, the

106

Chapter 4 Study 3

IRCC has an 86% match rate to the RPDB. Non-immigrants are those who remain unmatched to the IRCC database. A small proportion of the non-immigrant group may be immigrants who landed before 1985 [24].

4.3.2 Study population The study population included healthy singleton term siblings born in Ontario between April 1st,

2002 and March 31st, 2014 eligible for routine preventive care in Ontario from a pediatrician or family physician/general practitioner until 24.5 months of age. Included infants were those whose maternal country of birth was among the top 15 in Ontario (representing >90% of all births) and who had at least one eligible sibling of the opposite gender born in the study period.

Infants born at <37 weeks gestation, or weighing <2500 grams at birth, or those diagnosed with a complex chronic condition (e.g., asthma, cancer, major congenital malformations, among other conditions) [25] were excluded, as this may have influenced their experience of routine care. We excluded infants who received any primary care from community health centers (CHCs) because

CHCs do not submit billing data. Figure 4-1 presents a flowchart for the cohort formation.

SAS version 9.4 was used to perform all analyses (SAS Institute, Inc, Cary, NC). The study obtained ethics approval from the Research Ethics Board at the University of Toronto (Protocol reference #33799).

4.3.3 Variables

4.3.3.1 Outcomes Number of vaccinations by a family physician or pediatrician by 24 months of age. An immunization was measured by identifying the codes physicians use to bill the province for the administration of a vaccine and the number of units delivered on a given day. The total number

107

Chapter 4 Study 3 of vaccinations did not include unscheduled vaccinations occurring before six weeks. Infants were categorized as under-immunized if they had received less than the expected number of vaccinations publicly available and recommended in Ontario. Infants born in 2002-2003, 2004-

2009, 2010-11, and 2013 should have 5, 12, 11, and 10 vaccines respectively, against the availability in Ontario at the time (e.g., varied iterations and combinations of DPTP/Hib, MMR, pneumococcus, meningococcus and varicella vaccines).

Number of well-child visits by a family doctor or pediatrician by 24 months of age. These were operationalized by a set of core primary care fee codes and diagnostic codes and excluded immunization-only visits [26]. Infants were determined to have inadequate well-child check-ups if they had fewer than five check-ups by a family doctor or pediatrician (recommended at 2, 4, 6,

12, 15 (optional) and 18 months of age) in the first two years of life.

Vaccines or check-ups documented two weeks after 24 months were included as a buffer for billing delays or appointment scheduling. See Chapter 3 Appendices for further details on variable operationalization.

4.3.4 Exposures Maternal country of birth (MCOB) is recorded and entered at landing to Canada from notarized documents in the IRCC permanent resident database. If mothers did not have an immigration record, they were classified as Canadian-born. In this study, MCOB represents exposure to the health and gender-related norms of that country which may differentially affect the use of routine anticipatory care for sons and daughters.

Child gender. Gender norms, roles, and relations may differentially affect the receipt of routine, anticipatory health care for boys and girls within the family unit [9-12]. In this study, biological

108

Chapter 4 Study 3 sex documented at birth was described as gender, as the issue under study is gender-based discrimination.

4.3.5 Covariates Confounders were selected a priori and included variables that could vary between siblings [27].

They included maternal age at delivery of the index child (≤19, 20-34, ≥35 years of age), infant birth year [28], birth order (1st, 2nd, 3rd, 4th or more) [29], neighborhood income quintile

(1=lowest and 5=highest) and urban/rural residence (urban≤40 on Rurality Index of Ontario; rural ≥40) of the maternal place of residence at the birth of the index child [30] [31], and among immigrant mothers, time in Canada (≤5, 6-9, 10-14, 15+ years). Covariate data originated from the hospital record (maternal age, infant birth year, birth order), Canadian census data

(neighborhood income quintile and urban/rural residence), and IRCC documentation (landing date).

Statistics Canada’s Postal Code Conversion File links the mother’s postal code at delivery to generate values for the census-related variables. Deliveries before January 1st, 2004 were linked with the 2001 census, and deliveries afterward were linked to the 2006 census.

We examined maternal mother tongue as a potential effect modifier for the relationship between gender and routine preventive care among children of Indian-born mothers. Previous research has demonstrated variability in sex-selective pregnancy termination across language groups [18].

Mother tongue was recorded and documented in the IRCC at landing. We chose the three most commonly declared mother tongues to preserve sample size (Punjabi, Gujarati, Hindi, and

‘Other’).

109

Chapter 4 Study 3

4.3.6 Analysis Cross-tabulations and univariate procedures were used to obtain the unadjusted mean cumulative number and proportions of under-immunization and inadequate well-child visits for each gender within MCOB. Next, a fixed effects approach with conditional logistic regression stratified by

MCOB was used to estimate whether daughters have higher odds of adverse outcomes compared to sons within families. Fixed effects are useful for studying sibling differences by accounting for unobservable differences in maternal/family level variables [27, 32, 33]. Within-sibling variation is used to estimate the regression parameter for gender, and all stable characteristics of the family environment are controlled [32]—isolating the effect of child’s gender on anticipatory care outcomes within-families. Models were adjusted for maternal age, income quintile, rurality, birth year, birth order, and category of time spent in Canada (immigrant models only). A statistical test for interaction was performed between mother tongue*gender (significant at p≤0.2) to determine whether gender disparities are dependent on mother tongue for children of Indian-born mothers.

Variability in gender disparities analyses was also assessed by further stratifying by Indian mother tongue [11, 18].

In a sensitivity analysis, we looked at the interaction between gender and gender inequality in the country of origin at the time of migration as measured by the UN Gender Inequality Index (GII; low, medium, high, and very high gender inequality) on the outcomes. GII data was not available for all countries in all years. Next, we stratified models by GII quartiles [34].

110

Chapter 4 Study 3

4.4 Results

4.4.1 Demographic characteristics Figure 4-1 shows the cohort formation. Table 4-1 describes the characteristics of mothers and infants from each included MCOB. The total eligible population was 350,366 healthy, term, singleton infants, of 154,259 mothers from the 15 countries most commonly delivering in

Ontario. Countries represented included Canada, India, Pakistan, China, Philippines, Sri Lanka,

Jamaica, Afghanistan, Vietnam, Poland, Somalia, Iraq, USA, Guyana, and Iran.

Approximately 2/3 of women born in Pakistan (64.0%), USA (62.5%), and China (60.4%) had been in Canada for less than five years at the time of delivery, compared to only 14% of

Jamaican-born women.

4.4.2 Within-family gender disparities in routine preventive care Table 4-2 presents the prevalence of the outcomes by MCOB and gender, as well as unadjusted and adjusted within-family odds ratios and 95% CIs for under-immunization and inadequate well-child care for daughters compared to sons across MCOB strata.

No significant within-family gender differences were observed for immunization, although daughters whose mother was born in India had 1.08 times (95% CI 0.99, 1.17) the odds of under- immunization and 1.19 times (95% CI: 1.07, 1.33) the odds of inadequate check-ups than their male sibling. Girls in Afghani families had 27% greater odds of inadequate check-ups compared to their brothers (OR: 1.27, 95%CI: 1.01, 1.60), however that effect was no longer significant following covariate adjustment. We observed no significant within-family gender disparities for any other countries.

111

Chapter 4 Study 3

4.4.3 Effect modification by mother tongue among infants of Indian-born mothers The three most common mother tongues among Indian-born mothers were Punjabi (55.4%),

Gujarati (15.9%) and Hindi (9.4%). Among Indian-born mothers, statistical tests for interaction between mother tongue*gender were significant at p≤0.2 for both outcomes (under- immunization: chisq=4.79, p=0.19; inadequate well-child visits: chisq=5.27, p=0.15). Stratified analysis showed the relationship between gender and inadequate routine care was dependent on maternal mother tongue. Figure 4-2 shows significantly higher odds for sisters compared to brothers for inadequate well-child visits were present only for siblings in the Punjabi mother- tongue group (aOR:1.26, 95%CI: 1.08, 1.47), but not for Gujarati, Hindi or Other groups.

Daughters of mothers in the Hindi group had lower odds of under-immunization compared to their brothers (aOR: 0.73, 95%CI: 0.54, 0.98).

4.4.4 Results of sensitivity analyses We found a significant interaction between GII quartile and inadequate care outcomes (Appendix

4-2). Stratification by GII quartile demonstrated that odds of inadequate well-child visits were higher for daughters than sons among mothers who migrated from countries with very high gender inequality (aOR: 1.11, 95%CI: 1.02, 1.20). Other GII quartiles showed no significant differences.

4.5 Discussion 4.5.1 Summary of findings and interpretation To our knowledge, no studies have examined within-family gender disparities in early childhood routine care outcomes by maternal country of birth, which is essential given documented gender disparities in many source countries of immigrants to Canada. We did not find evidence of gender disparities in under-immunization or inadequate well-child visits for most MCOB, except

112

Chapter 4 Study 3 among Indo-Canadian families where sisters had significantly higher odds of inadequate well- child visits compared to their brothers.

Our study contrasts against earlier work that did not find gender disparities in vaccinations among British South Asian children [20]. By using a larger sample size and fixed effects analysis

[27], we were able to estimate family-held gender biases adversely affecting health care for daughters compared to sons within families. We found that gender disparities within Indo-

Canadian families were dependent on the Punjabi mother tongue. This finding is consistent with earlier studies of sex-selective terminations in the Indo-Canadian community [18] as well as research from the North-West of India where Punjabi is a dominant language [35, 36]. Research in this region describes beliefs about economic benefits of sons over daughters, including the provision of old age support, higher-paid employment, patrilineal kinship systems, and avoiding the high cost of dowry or marriage payments [5, 12, 35, 37]. For some families, son preference may manifest primarily through sex selection but also through health-care neglect [8]. It is possible for similar mechanisms related to sex selection to influence gender equity in infant health care. In unconditional models, we found female disadvantage among children of among children of Afghanistan-born mothers for inadequate well-child visits. In adjusted models the effect estimate did not change substantially, but precision suffered. Therefore, it is plausible that the 23% increased odds of inadequate females is relevant to clinical or public health practice, and would be statistically significant with a larger sample size, such as in the case of India [41].

Sensitivity analyses revealed significant within-family gender disparities only among infants of mothers born in countries with very high gender inequality at the time of migration to Canada.

This finding further suggests that daughters may be at risk of inadequate preventive health care

113

Chapter 4 Study 3 for some families originating from countries with high gender inequality; however, the large

Indian-born maternal population may be driving some of these results.

4.5.2 Strengths To our knowledge, this was the first retrospective population-based cohort to examine gender disparities across multiple routine preventive care outcomes among children of immigrants.

Second, due to the substantial diversity of Ontario mothers, we were able to examine gender disparities across many MCOB [38]. Third, this study is the first to examine the issue of gender bias occurring within families that may be affecting the health of children of immigrants and non-immigrants. Finally, the use of: 1) official government immigration data to identify immigration factors; and 2) population-based administrative health data help strengthen both the internal and external validity of the study.

4.5.3 Limitations Immigration data is only available for those arriving in Canada after 1985, so those arriving prior were misclassified as non-immigrants. Therefore, the Canadian-born group may be heterogeneous with health and gender beliefs related to their ancestral immigrant group.

However, given an extended duration of residence, beliefs and health practices may be closer to

Canadian norms [39]. As well, this population is likely small relative to the Canadian-born population. Different parental countries of birth may have introduced bias toward the null effect.

Mother tongue may not necessarily represent a particular geographic region of India nor her most commonly spoken language around the time of the index delivery [18], introducing possible heterogeneity.

Finally, census-derived variables are measured infrequently, and may result in misclassification, contributing to residual confounding. Using a fixed-effects analytic approach helped to control

114

Chapter 4 Study 3 for within-family unobservable factors [27, 32, 33, 40]. It is also important to note that immigrants are not representative of the source population, which may explain why we did not find significant effects for most MCOB, even among those where gender inequity is high [35,

37].

4.5.4 Implications Son preference appears to persist beyond family planning and adversely affects the wellbeing of daughters whose mothers migrate from countries with high gender inequality, including India.

This work may help health care providers attend to children needing additional preventive care.

Future directions include verifying the mechanisms behind son preference in child health care, as well as exploring potential gender-based adversity through other areas of child well-being.

Community-led interventions addressing son preference and the well-being of daughters may be helpful in improving gender equity in health care.

115

Chapter 4 Study 3

Table 4-1 Descriptive characteristics of included mother-infant sibling sets with at least one boy and one girl, among the 15 countries with the greatest share of births in Ontario, April 2002- March 2013

Maternal Country of Birth N infants=350,336 Sri Canada India Pakistan China Philippines Jamaica Afghanistan Vietnam Somalia Poland Iraq USA Guyana Iran 154,259 mothers Lanka 290,009 12,356 9150 7566 5654 6169 3057 2684 2424 2382 2340 2030 1828 1626 1061 N (%) (82.8) (3.5) (2.6) (2.2) (1.6) (1.8) (0.9) (0.8) (0.7) (0.7) (0.7) (0.6) (0.5) (0.5) (0.3) Maternal Age % ≤19 3.80 0.4 0.8 0.2 1.1 0.3 5.3 1.2 0.5 1.1 0.8 1.1 1.0 2.5 0.5 20-34 80.1 90.0 87.5 75.5 69.3 86.4 79.8 85.2 80.9 74.0 83.9 81.8 75.7 82.1 68.9 35+ 16.1 9.6 11.7 24.3 29.5 13.2 14.8 13.4 18.6 24.7 15.3 17.1 23.2 15.4 30.6 Female % 49.8 51.0 50.1 50.3 49.8 49.9 49.5 50.5 50.1 50.0 49.5 50.0 50.1 49.9 49.9 Birth Order % First 36.8 41.2 35.8 45.0 39.5 39.7 30.6 31.7 39.2 18.5 40.2 32.2 34.3 35.7 46.6 Second 41.1 44.6 39.2 46.3 43.4 42.2 38.5 37.1 43.1 23.7 43.5 37.9 37.6 42.1 48.0 3rd 15.5 11.6 19.1 7.7 13.7 15.4 18.9 20.3 13.6 20.9 13.2 21.0 16.0 15.6 4.6 4th or greater 6.7 2.6 6.0 1.0 3.4 2.7 12.0 10.8 4.2 37.0 3.1 8.9 12.1 6.6 0.9 Time since landing % <5 years - 57.8 64.0 60.4 43.8 55.5 13.8 45.5 40.0 32.7 16.1 53.6 62.5 21.6 42.6 5-9 years - 28.8 25.9 31.0 25.5 23.2 23.0 30.1 24.1 30.1 13.6 26.1 18.8 26.2 28.0 10-14 years - 9.2 7.6 6.9 17.9 13.2 29.7 15.3 17.5 25.4 26.0 13.8 8.7 24.8 15.3 15+ years - 4.3 2.6 1.7 12.7 8.1 33.5 9.2 18.4 11.9 44.2 6.5 10.1 27.4 14.1 Neighborhood

Income Quintile 1: Lowest 17.6 28.6 39.9 26.4 32.3 36.7 44.6 50.4 31.9 70.0 13.0 43.0 13.2 36.1 17.3 2 18.5 27.5 22.8 30.2 24.1 29.0 23.8 17.6 25.0 16.4 20.7 21.4 18.4 23.3 11.8 3 20.6 24.3 18.2 19.5 19.7 20.5 18.7 13.6 22.7 7.0 21.2 16.6 18.7 24.5 20.0 4 23.0 14.0 14.1 15.3 15.4 10.3 8.6 12.3 14.5 4.9 26.5 13.0 23.4 11.6 31.5 5: Highest 19.6 5.7 5.0 8.3 8.5 3.4 4.2 6.0 5.8 1.5 18.6 5.7 26.2 4.6 19.1 Lives in Rural Area 14.7 0.3 0.3 0.3 1.0 0.0 0.2 0.3 0.2 0.0 2.0 0.2 10.7 0.4 0.2 Gender Inequality

Quartile at landing Low (<0.280) 94.0 0 82.0 0 0 0 0 0 0 0 100 0 1 0 0 Med (0.280-0.512) 0 0 18.0 0 100 100 46.6 0 100 0 0 0 99 69.0 0 High (0.512-0.674) 0 26.5 0 30.2 0 0 53.4 0 0 0 0 100 0 31.0 77.9 Very High (>0.674) 0 73.5 0 73.5 0 0 0 100 0 0 0 0 0 0 22.1 Unavailable 6.0 0 0 0 0 0 0 0 0 100 0 0 0 0 0

116

Chapter 4 Study 3 Table 4-2 Within-family Gender Disparities in Routine Preventive Care Outcomes Stratified by Maternal Country of Birth, among all eligible siblings sets in Ontario 2002-2013

Under-Immunization Inadequate Well-child Visits % F:M OR (95% CI) F:M aOR (95% CI) % F:M OR (95% CI) F:M aOR (95% CI) Maternal Country F M F M of Birth Canada 44.6 44.4 0.99 (0.97, 1.01) 0.99 (0.97, 1.00) 16.3 17.0 0.94 (0.91, 0.96) † 0.96 (0.92, 1.00) India 38.1 36.9 1.05 (0.97, 1.14) 1.08 (0.99, 1.17) 18.8 16.9 1.17 (1.05, 1.30) † 1.19 (1.07, 1.33) † Pakistan 41.3 42.7 0.95 (0.97, 1.05) 0.96 (0.87, 1.05) 21.9 22.7 0.94 (0.84, 1.05) 0.95 (0.85, 1.07) China 30.2 28.9 1.09 (0.97, 1.22) 1.07 (0.95, 1.20) 19.9 19.9 1.01 (0.88, 1.16) 1.02 (0.89, 1.17) Philippines 30.5 31.1 0.95 (0.84, 1.08) 0.98 (0.89, 1.11) 13.4 14.6 0.88 (0.74, 1.05) 0.89 (0.74, 1.07) Sri Lanka 34.2 34.3 0.99 (0.88, 1.12) 1.00 (0.88, 1.12) 16.5 17.2 0.94 (0.81, 1.10) 0.92 (0.78, 1.08) Jamaica 45.2 44.6 1.03 (0.88, 1.20) 0.98 (0.83, 1.16) 26.9 25.4 0.93 (0.77, 1.12) 0.9 (0.73, 1.10) Afghanistan 38.3 40.1 0.89 (0.75, 1.06) 0.90 (0.75, 1.07) 18.8 16.1 1.27 (1.01, 1.60) † 1.23 (0.96, 1.56) Vietnam 27.9 27.8 1.01 (0.82, 1.24) 1.02 (0.82, 1.26) 10.9 10.5 0.9 (0.66, 1.23) 0.82 (0.58, 1.16) Somalia 58.2 53.1 0.79 (0.67, 0.95) 0.84 (0.69, 1.03) 44.5 47.8 0.87 (0.72, 1.05) 0.99 (0.76, 1.50) Poland 45.1 43.5 1.08 (0.90, 1.30) 1.10 (0.91, 1.35) 16.1 15.6 1.07 (0.83, 1.39) 1.02 (0.75, 1.38) Iraq 37.3 37.0 0.99 (0.81, 1.22) 1.03 (0.83, 1.28) 18.0 17.7 0.97 (0.74, 1.26) 0.99 (0.75, 1.31) USA 45.5 46.8 0.96 (0.78, 1.18) 0.93 (0.75, 1.15) 20.3 19.2 1.17 (0.91, 1.51) 1.2 (0.87, 1.66) Guyana 39.1 38.3 1.03 (0.83, 1.29) 1.07 (0.84, 1.35) 17.1 20.2 0.81 (0.61, 1.07) 0.83 (0.61, 1.14) Iran 26.5 28.0 0.94 (0.67,1.28) 0.95 (0.67, 1.34) 13.7 11.0 0.76 (0.50, 1.16) 0.86 (0.51, 1.45) Adjustment was for maternal age, income quintile, rurality, birth year, birth order, and category of time spent in Canada (immigrant models only). OR odds ratio, aOR adjusted odds ratio, 95% CI 95% Confidence Interval †indicates statistical significance at p<0.05. Reference group= males

117

Chapter 4 Study 3

Figure 4-1 Flow Chart of Cohort Formation

118

Chapter 4 Study 3

Figure 4-2 Within-family Gender Disparities (Female:Male Unadjusted ● and Adjusted ■ Odds Ratios) in Under- Immunization and Inadequate Well-Child Visits among Children of Indian-born Mothers, Stratified by Mother Tongue. Adjustment was for maternal age, income quintile, rurality, birth year, birth order, and category of time spent in Canada.

119

Chapter 4 Study 3 4.6 Appendices

Appendix 4-1 Number and proportion of excluded infant siblings by maternal country of birth and gender

Maternal country of Birth Canada India China Pakistan Philip. Sri Lanka Jamaica Poland Vietnam Afghanist. Guyana USA Iraq Iran Somalia Number of opposite gender sibling births 444,622 18,205 10,621 13,350 8103 9585 4725 3304 3279 3634 2555 2782 2813 1521 3621 Number of infants with at least one exclusion criteria (%) 109,268 551 670 609 680 541 313 874 3821 (21.9) 1996 (18.8) 2755 (20.6) 1596 (19.7) 2255 (23.5) 1121 (23.7) 635 (19.2) (24.6) (16.8) (18.4) (23.8) (24.4) (19.2) (20.6) (24.1) Number of female siblings excluded (%) 53,528 268 339 297 335 273 145 414 1913 (20.6) 977 (18.3) 1315 (20.1) 780 (19.4) 1078 (22.5) 556 (23.6) 319 (19.6) (24.2) (16.3) (18.5) (23.2) (24.2) (19.3) (19.1) (22.9) Number of male siblings excluded (%) 55,740 283 331 312 345 268 168 460 1908 (21.4) 1019 (19.3) 1410 (21.2) 816 (20.0) 1177 (24.5) 565 (23.8) 316 (18.9) (24.9) (17.4) (18.4) (24.4) (24.7) (19.1) (22.1) (25.4) Final number of infants where both siblings meet all inclusion criteria 290,009 12,356 7566 9150 5654 6169 3057 2340 2424 2684 1626 1828 2030 1061 2382

120

Chapter 4 Study 3

Appendix 4-2 Sensitivity Analyses: Odds of routine preventive care for Ontario-born daughters compared to sons across levels of gender inequality in the maternal country of birth

Under- Inadequate Well- Immunization Child Visits UN gender inequality quartile in MCOB at N(%) aOR (95% CI) aOR (95% CI) landing date (index value) Low (<0.280) 285, 549 (81.50) 1.05 (0.95, 1.17) 1.03 (0.91, 1.18) Med (0.280-0.512) 20,009 (5.71) 1.00 (0.94, 1.07) 0.93 (0.85, 1.02) High (0.512-0.674) 10,865 (3.10) 1.03 (0.95, 1.13) 0.97 (0.86, 1.08) Very High (>0.674) 18,549 (5.29) 1.01 (0.94, 1.08) 1.11 (1.02, 1.20) † Missing 15,364 (4.39) 0.84 (0.69, 1.03) 0.99 (0.80, 1.22) Adjusted Chi Adjusted Chi sq= GII quartile*gender sq=7.43, p=0.12† 12.45, p=0.01† Within-family gender disparities stratified by GII quartile: Stratification by GII quintile demonstrated that inadequate well-child check-ups were higher for girls than boys among mothers who migrated from countries ranked very high on the GII, no differences were seen for other GII quartiles. *Adjusted for income quintile, birth order, maternal age, duration of residence, rurality aOR adjusted odds ratio GII data not available for Somalia, or some years for Canada †indicates statistical significance at p<0.05.

121

Chapter 4 Study 3

4.7 References

1. Condon, L. and S. McClean, Maintaining pre-school children's health and wellbeing in the UK: a qualitative study of the views of migrant parents. Journal of Public Health, 2016. 39(3): p. 455-463. 2. Brown, K.F., et al., Factors underlying parental decisions about combination childhood vaccinations including MMR: a systematic review. Vaccine, 2010. 28(26): p. 4235-4248. 3. Borràs, E., et al., Parental knowledge of paediatric vaccination. BMC Public Health, 2009. 9(1): p. 154. 4. Lhila, A. and K.I. Simon, Prenatal health investment decisions: does the child's sex matter? Demography, 2008. 45(4): p. 885-905. 5. Puri, S., et al., “There is such a thing as too many daughters, but not too many sons”: A qualitative study of son preference and fetal sex selection among Indian immigrants in the United States. Social science & medicine, 2011. 72(7): p. 1169-1176. 6. Ebenstein, A., H. Li, and L. Meng, The impact of ultrasound technology on the status of women in China, in Hebrew University and Tsinghua University Working Paper. 2010. 7. Bharadwaj, P., G.B. Dahl, and K. Sheth, Gender discrimination in the family. The economics of the family: How the household affects markets and economic growth, 2014. 2: p. 237. 8. Khera, R., et al., Gender bias in child care and child health: global patterns. Archives of disease in childhood, 2013: p. archdischild-2013-303889. 9. Barcellos, S.H., L.S. Carvalho, and A. Lleras-Muney, Child gender and parental investments in India: Are boys and girls treated differently? American Economic Journal: Applied Economics, 2014. 6(1): p. 157-89. 10. Pande, R.P., Selective gender differences in childhood nutrition and immunization in rural India: the role of siblings. Demography, 2003. 40(3): p. 395-418. 11. Pande, R.P. and A.S. Yazbeck, What's in a country average? Wealth, gender, and regional inequalities in immunization in India. Social Science & Medicine, 2003. 57(11): p. 2075-2088. 12. Jayachandran, S. and I. Kuziemko, Why do mothers breastfeed girls less than boys? Evidence and implications for child health in India. The Quarterly journal of economics, 2011. 126(3): p. 1485-1538. 13. Fikree, F.F. and O. Pasha, Role of gender in health disparity: the South Asian context. BMJ: British Medical Journal, 2004. 328(7443): p. 823. 14. Ray, J.G., D.A. Henry, and M.L. Urquia, Sex ratios among Canadian liveborn infants of mothers from different countries. Canadian Medical Association Journal, 2012: p. cmaj. 120165. 122

Chapter 4 Study 3

15. Singh, N., et al., Different sex ratios of children born to Indian and Pakistani immigrants in Norway. BMC pregnancy and childbirth, 2010. 10(1): p. 40. 16. Urquia, M.L., et al., Sex ratios at birth after induced abortion. Canadian Medical Association Journal, 2016. 188(9): p. E181-E190. 17. Urquia, M.L., et al., Variations in male-female infant ratios among births to Canadian- and Indian-born mothers, 1990-2011: a population-based register study. CMAJ open, 2016. 4(2): p. E116. 18. Brar, A., et al., Sex Ratios at Birth Among Indian Immigrant Subgroups According to Time Spent in Canada. Journal of Obstetrics and Gynaecology Canada, 2017. 39(6): p. 459-464. e2. 19. Srinivasan, S., Transnationally relocated? Sex selection among Punjabis in Canada. Canadian Journal of Development Studies/Revue canadienne d'études du développement, 2018: p. 1-18. 20. Martineau, A., M. White, and R. Bhopal, No sex differences in immunisation rates of British south Asian children: the effect of migration? BMJ, 1997. 314(7081): p. 642. 21. Pulver, A., et al., A scoping review of female disadvantage in health care use among very young children of immigrant families. Social Science & Medicine, 2016. 152: p. 50-60. 22. Vertovec, S., Super-diversity and its implications. Ethnic and racial studies, 2007. 30(6): p. 1024-1054. 23. Wanigaratne, S., et al., The influence of refugee status and secondary migration on preterm birth. J Epidemiol Community Health, 2016. 70(6): p. 622-628. 24. Chiu, M., et al., Describing the linkages of the immigration, refugees and citizenship Canada permanent resident data and vital statistics death registry to Ontario’s administrative health database. BMC medical informatics and decision making, 2016. 16(1): p. 135. 25. Feudtner, C., et al., Pediatric complex chronic conditions classification system version 2: updated for ICD-10 and complex medical technology dependence and transplantation. BMC pediatrics, 2014. 14(1): p. 199. 26. Guttmann, A., et al., Primary care physician supply and children's health care use, access, and outcomes: findings from Canada. Pediatrics, 2010. 125(6): p. 1119-1126. 27. Allison, P.D., Fixed effects regression methods for longitudinal data using SAS. 2005: Sas Institute. 28. Schwartz, K.L., et al., Validation of infant immunization billing codes in administrative data. Human vaccines & immunotherapeutics, 2015. 11(7): p. 1840-1847. 29. Hawken, S., et al., Association between birth order and emergency room visits and acute hospital admissions following pediatric vaccination: a self-controlled study. PloS one, 2013. 8(12): p. e81070.

123

Chapter 4 Study 3

30. Kralj, B., Measuring “rurality” for purposes of health-care planning: an empirical measure for Ontario. Ont Med Rev, 2000. 67(9): p. 33-52. 31. Gushulak, B.D., et al., Migration and health in Canada: health in the global village. Canadian Medical Association Journal, 2011. 183(12): p. E952-E958. 32. D'onofrio, B.M., et al., Familial confounding of the association between maternal smoking during pregnancy and offspring criminality: a population-based study in Sweden. Archives of general psychiatry, 2010. 67(5): p. 529-538. 33. Lahey, B.B. and B.M. D’Onofrio, All in the family: Comparing siblings to test causal hypotheses regarding environmental influences on behavior. Current directions in psychological science, 2010. 19(5): p. 319-323. 34. Gaye, A., et al., Measuring key disparities in human development: The gender inequality index. Human Development Research Paper, 2010. 46: p. 1-37. 35. Das Gupta, M., et al., Why is son preference so persistent in East and South Asia? A cross-country study of China, India and the Republic of Korea. The Journal of Development Studies, 2003. 40(2): p. 153-187. 36. Gupta, M.D., Selective discrimination against female children in rural Punjab, India. Population and development review, 1987: p. 77-100. 37. Jayachandran, S., The roots of gender inequality in developing countries. economics, 2015. 7(1): p. 63-88. 38. Pavlish, C.L., S. Noor, and J. Brandt, Somali immigrant women and the American health care system: discordant beliefs, divergent expectations, and silent worries. Social Science & Medicine, 2010. 71(2): p. 353-361. 39. Reitz, J.G., M.B. Phan, and R. Banerjee, Gender equity in Canada's newly growing religious minorities. Ethnic and Racial Studies, 2015. 38(5): p. 681-699. 40. Durbin, A., et al., Examining the relationship between neighbourhood deprivation and mental health service use of immigrants in Ontario, Canada: a cross-sectional study. BMJ open, 2015. 5(3): p. e006690. 41. Hackshaw, A., & Kirkwood, A. Interpreting and reporting clinical trials with results of borderline significance. BMJ, 2011 343: d3340.

124

Chapter 5 Study 4

Chapter 5, Study 4: Differences in Early Childhood Maltreatment by Maternal Birthplace and Child Gender

125

Chapter 5 Study 4

5.1 Abstract

Background

There exist cultural differences in child-rearing practices worldwide and across immigrant groups. It is unknown if maltreatment identified in health-care settings in early childhood varies by parental immigration status.

Objective

Within a multicultural society, where universal health care is available, we examined patterns of health-care identified early childhood maltreatment by 1) maternal birthplace, and 2) child gender across maternal birthplace.

Methods

This retrospective population-based cohort study comprised 1,240,946 children born in Ontario,

Canada between 2002 and 2012, and followed from birth to age 5 using administrative data.

Modified Poisson regression was used to estimate rate ratios (RR) for maltreatment – physical abuse or neglect – among the children of immigrant vs. non-immigrant mothers. Conditional logistic regression was used to estimate further the odds of maltreatment comparing a daughter vs. son of the same mother.

Results

Maltreatment rates were 36% lower (RR=0.64, 95%CI 0.61-0.66) lower among children of immigrant mothers (10 per 1000) than those of non-immigrant mothers (16 per 1000).

Maltreatment rates were 28-48% lower among children of maternal immigrant groups relative to that among Canadian-born mothers, except children of Caribbean-born mothers (16 per 1000).

126

Chapter 5 Study 4

No significant differences were seen between daughters and sons in the odds of early childhood maltreatment by maternal birthplace.

Conclusions

Early childhood maltreatment is highest among a select group of children of Canadian-, and

Caribbean-born mothers. Maltreatment did not differ between daughters and sons of the same mother. These data may help focus strategies aimed at reducing maltreatment among vulnerable groups.

127

Chapter 5 Study 4

5.2 Introduction Although child maltreatment is a global phenomenon,[1, 2] some ethnic and racial groups are disproportionately represented in child-welfare systems.[3-7] Understanding culturally-based parenting differences in health care and child protective services has therefore gained importance.[3-13] However, prior studies of child maltreatment have not sufficiently distinguished between ethnic groups, have been limited to one specific group,[11] or have combined recent immigrants with third-generation families.[9] Furthermore, few studies have explored the contribution of both parental birthplace and immigration to childhood maltreatment.

Existing studies relied on self-reported maltreatment data.[10, 14]

Immigration-related differences in maltreatment may also vary by a child’s gender, depending, for example, on a home country’s societal attitudes towards men and women. Research shows that unique structural factors and individually held beliefs adversely affect the health and wellbeing of girls from top source countries of immigrants to Canada, including India, Pakistan, and China.[15-18] Among immigrant populations female disadvantage may exist throughout the life course, manifesting through fetal sex selection,[19-22] female genital mutilation[23] and domestic violence.[24-27] Overall, evidence remains inconclusive about the persistence of gender inequalities related to child maltreatment, as gender inequalities in maltreatment may vary by maternal birthplace in a high-immigration context.[28-32] This population-based study was undertaken to identify: i) whether rates of early childhood maltreatment differ by maternal birthplace; and ii) whether the gender of the child further alters the risk of maltreatment.

128

Chapter 5 Study 4

5.3 Methods

5.3.1 Design, participants, and setting We conducted a population-based cohort study completed in Ontario, Canada, where universal health care is provided to all children and adults under the Ontario Health Insurance Programme

(OHIP). The study sample included all children born in Ontario, between April 1st, 2002 and

March 31st, 2012, and who were eligible for OHIP from birth until 60 months of age. To ensure complete follow-up for children from birth to age five years, they were excluded if they left the province, experienced a death not resulting from maltreatment, or had no contact with the health care system for a consecutive period of 18 months suggesting they may have left the province.

We also excluded children with missing covariate data. A sub-cohort of opposite-gender sibling sets were selected from the primary cohort (Figure 5-1) to explore whether a child's gender influences the risk of maltreatment.

5.3.2 Data sources We analyzed linked population-based administrative databases housed at ICES in Toronto,

Canada. These datasets are linked using unique encoded identifiers.

The Registered Persons Database (RPDB) contains information on birth date, sex, and postal code. Canadian Census data (2001 and 2006) provide neighborhood information at the level of a dissemination area, such as residential income quintile. Children born alive in Ontario between

April 1st, 2002 and March 31st, 2012 were identified via hospitalization-related births in the

Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD), capturing

98% of all births in Ontario. The National Ambulatory Care Reporting System and CIHI-DAD each use International Classification of Diseases 10th Revision-Canada (ICD-10CA) diagnostic codes for all emergency department visits, and inpatient hospitalizations, respectively. Data from

129

Chapter 5 Study 4

Ontario’s Vital Statistics database provides details on out-of-hospital deaths. The Ontario portion of the federally maintained Immigration, Refugees, and Citizenship Canada (IRCC) Permanent

Resident Database was used to provide information about maternal birthplace and date of arrival to Canada. The IRCC database includes all landed immigrants who obtained permanent residence from January 1, 1985, to December 31, 2012. The overall linkage rate for the IRCC database to the RPDB is 86% [20, 21, 33, 34].

5.3.3 Study variables

5.3.3.1 Exposures Maternal birthplace, recorded during the immigration application process from notarized copies of original documents, is entered in the IRCC Permanent Resident Database. If mothers did not have an immigration record, they were classified as non-immigrants and assumed to be

Canadian-born. A small proportion of immigrants who migrated to Ontario before 1985 or moved there from another province may be incorrectly classified as non-immigrants, but this would not be expected to meaningfully bias the outcome estimates within the Canadian-born group. Immigrant maternal birthplace included the four most common countries of birth for mothers who gave birth in Ontario, with the others categorized by world region of birth: India,

China, Pakistan, Philippines, Other South Asia, Other East Asia/Pacific, Latin America,

Caribbean, North Africa/Middle East, Other Western countries, Eastern Europe, and Sub-

Saharan Africa.

Child gender. In this study, the biological sex of the child at birth was denoted as “gender."

5.3.3.2 Outcome The definition for child maltreatment was restricted to events of physical abuse or neglect captured with certainty within the health care system [35]. Maltreatment events were counted

130

Chapter 5 Study 4 from fatal or non-fatal injuries resulting in emergency department visits or hospitalizations, or deaths occurring outside of hospital – each from the birth date of the child to 60 months after that. Outcomes were captured by either ICD-10CA codes at the time of an emergency department visit or hospital admission, or by the cause of death code among out-of-hospital deaths (Appendix 5-1).

Population-based ascertainment of maltreatment-related events identified in health-care settings has been previously operationalized and compared internationally by Gilbert and colleagues, and published in the Lancet [35]. Maltreatment events identified in health care administrative data consist of four subcategories of maltreatment-related events, in descending order of specificity

[35]. These include 1. maltreatment syndrome (physical abuse or neglect as the cause of injury),

2. assault (violence by caregivers (physical abuse), or others, which may be due to inadequate supervision (neglect)), 3. undetermined cause (explicit uncertainty about the cause of injury, which may reflect physical abuse or neglect), and 4. adverse social circumstances (physician concern about parenting, home environment, or broader welfare concerns associated with the injury). Further details are found in Appendix 5-1.

5.3.3.3 Covariates Potential confounders were selected a priori and included variables associated with immigration status and child maltreatment. Data on maternal age at delivery of the index child (≤19, 20-34,

≥35 years of age) [36], plurality (singleton or multiple births) [37, 38], and childhood complex chronic conditions (yes/no) [39, 40] was sourced from hospital records. Neighborhood income quintile (1 = lowest to 5 = highest) and urban/rural residence (urban ≤ 40 on Rurality Index of

Ontario; rural ≥40) [41] was ascertained from Canadian census data using the closest census year

131

Chapter 5 Study 4 to the childbirth. Among immigrant mothers, time since landing in Canada at the delivery of the index child (≤ 5, 6-9, 10-14, 15+ years) was obtained from IRCC documentation.

5.3.4 Statistical analysis To assess differences according to maternal birthplace, we first ran cross-tabulations to obtain the unadjusted 5-year cumulative incidence of early childhood maltreatment/1000 children for each maternal birthplace and for boys and girls within each maternal birthplace stratum. We then used modified Poisson regression to estimate crude and adjusted rate ratios (RR, aRR) and corresponding 95% confidence intervals (95%CI) for maltreatment among children of immigrants compared to those of Canadian-born mothers, as well as by specific maternal birthplace compared to Canada [42, 43]. Models assessing maternal birthplace differences adjusted for infant gender, maternal age group, plurality, birth order, neighborhood income quintile, urban/rural residence, childhood complex chronic conditions, and accounted for the possible birth of more than one child per mother [42, 43] We also compared sub-categories of health system-identified maltreatment that comprised the primary outcome variable between children of immigrants and non-immigrants using unadjusted rate ratios.

We then used a fixed effects approach with conditional logistic regression to estimate whether daughters have higher odds of maltreatment compared to sons in the same family, both overall and per maternal birthplace (OR, aOR). Fixed effects are useful for studying sibling differences by accounting for unobservable differences in maternal/family level variables [44-46]. Within- family variation is used to estimate the regression parameter for gender, and all stable characteristics of the family environment are controlled [44-46]. Models assessing within-family gender differences were adjusted for maternal age at delivery, income quintile, rurality

(Canadian-born stratum only), birth order, and category of time in Canada since obtaining

132

Chapter 5 Study 4 permanent residence (immigrant maternal birthplace strata only). The fixed effects approach controls other non-varying immigration characteristics.

5.4 Results

5.4.1 Descriptive Characteristics There were 1,315,186 births in Ontario in the study period. Of these children, we excluded

28,194 due to loss of provincial health insurance from outmigration. We excluded an additional

38,500 because they had no health system contact for 18 months, suggesting they may have left the province without informing the provincial health insurance plan. Finally, 6,145 more children were removed due to missing covariate data, resulting in a final study population of 1,240,946

(94.4% of births) (Figure 5-1). One-quarter of children were born to immigrant mothers (Table

5-1). Children of immigrant mothers were more often born into the lowest income quintile and younger than children of Canadian-born mothers.

5.4.2 Differences in Early Childhood Maltreatment Rate by Maternal Region of Birth Early childhood maltreatment was significantly higher among children of Canadian-born mothers (16 per 1000 children) compared to immigrants (10 per 1000 children, p<0.001) (Table

2). The 5-year rate of child maltreatment varied significantly according to maternal birthplace

(Table 5-2). It was highest among children of Canadian-, Sub-Saharan African- (12 per 1000) and Caribbean- (16 per 1000) born mothers and lowest among children whose mothers were born in India (8 per 1000) and China (8 per 1000). Compared to children of Canadian-born mothers, the rates among children of Caribbean women did not differ. Children of all other immigrant

133

Chapter 5 Study 4 mothers were 28-48% less likely to have experienced early childhood maltreatment than children of Canadian-born mothers. Among children with maltreatment-related health care documentation, the subcategories of maltreatment syndrome, adverse social circumstances, and undetermined cause of injury were significantly more common among children of Canadian-born mothers than those of immigrants (Table 5-3). There was no difference in assaults. Due to small numbers, we could not further disaggregate analyses by the frequency or type of maltreatment by maternal birthplace. We found no evidence of within-family gender differences in maltreatment among siblings (n=422,880, 23% to immigrant mothers, Table 5-4).

5.5 Discussion Our findings indicate lower rates of health system-identified maltreatment for children of almost all groups of immigrant mothers compared to children of Canadian-born mothers. Children of

Caribbean-born mothers did not differ from the Canadian group. There was no evidence of gender differences in child maltreatment for any maternal birthplace. Our study addresses a significant gap where maternal immigration background and child gender intersect. Before accounting for socioeconomic differences, Black children are over-represented in US and

Canadian child welfare systems [4, 47]. However, maternal immigration background has rarely been accounted for in previous analyses [4, 7, 47]. Involvement with child protective services may arise from increased use of physical discipline, lower adult supervision, or assigning older children to care for younger siblings [7], and may be compounded by racially-based surveillance and criminalization [4, 7].

Adjustment for covariates did not account for reduced risk among the remaining immigrant groups, suggesting that disproportionate representation across groups may not be entirely explained by socioeconomic factors. Future research into the contribution of sociodemographic

134

Chapter 5 Study 4 factors on child maltreatment among immigrants and non-immigrants would shed light on this grey area.

We observed lower maltreatment rates among children of mothers born in all Asian regions, consistent with reports of under-representation of Asian-Canadian children [6]. Again, researchers did not consider parental immigration backgrounds. Child maltreatment appears to be associated not only with ethnicity but with immigration specifically. US Latino immigrant families are at lower risk for child maltreatment than Latino non-immigrants. Immigrant communities with high proportions of two-parent families and low rates of adolescent pregnancy and substance use may buffer challenges related to poverty, low neighborhood safety, and limited access to social services [9, 11].

In contrast, European research has shown higher rates of child maltreatment among immigrant groups, although these studies ascertained maltreatment through adolescent self-report [10, 13].

Immigrant communities may be more heterogeneous and dispersed in Europe compared to the

US and Canada, where cohesion facilitates stronger community, ethnic and familial affiliations, helping to buffer immigration-related challenges [13]. Moreover, Canada has a highly selective immigration process that strongly favors well-educated immigrants proficient in the host language [48].

We found no evidence of gender differences in overall maltreatment in any maternal birthplace.

Gender differences may exist for specific forms of maltreatment but not for others. While boys and girls have higher rates of physical abuse and sexual abuse respectively [38, 49], it is unknown whether gender differences exist for siblings affected by maltreatment. Although maternal birthplace may not impact gender differences in child maltreatment, it may impact the propensity for family violence in general.

135

Chapter 5 Study 4

We would like to highlight four study strengths. First is the population-based cohort providing strong representativeness. Second is the 5-year follow-up period thereby limiting recall issues seen in self-report studies. A third strength is the fixed effects analytic approach to isolate the effect of child gender on maltreatment. Finally, maltreatment documented in health care settings may indicate a high degree of certainty that maltreatment has occurred, because it captures the most severe forms resulting in injuries requiring medical attention.

Reliable measurement of child maltreatment is not straightforward [50]. Health system-identified maltreatment likely underestimates the real burden and does not capture some common forms

(e.g., witness to intimate partner violence) [51, 54-55]. Maltreatment reports from child protective services also tend to underestimate maltreatment because not all maltreated children interact with child protection services [50, 52]. However, child protection agencies in Canada recently reported 53.3 maltreatment investigations per 1000 children in 2013, 34% which were substantiated, bolstering the face validity of the current study’s observed rates [53]. In contrast, rates generated from self-report studies range from 5-35%, though self-report is not feasible in very early childhood.

There are five other limitations. First, we were unable to examine differences in specific forms of maltreatment due to their relative rarity. Second, we excluded children with incomplete follow- up time who may have been at risk of maltreatment. Similarly, findings may not be generalizable to children born outside of Canada due to incomplete follow-up. Third, issues may arise in clinical settings that are not captured well by the ICD-10-CA or billing codes resulting in undercounting of maltreatment. Fourth, we could not adjust for some immigration characteristics such as duration of residence and refugee status due to the absence of variation in these covariates among Canadian-born mothers. Moreover, the diagnoses of complex chronic

136

Chapter 5 Study 4 conditions may itself be related to the maltreatment admission, perhaps creating some residual confounding.

5.5.1 Conclusion Maternal birthplace influences the rate of early childhood maltreatment in a high immigration setting, but it does not further vary according to child gender. The protective effect of maternal immigration background provides more nuanced evidence surrounding racial and ethnic differences in child maltreatment. Efforts to further understand how maternal immigration protects against child maltreatment may provide clues on prevention strategies for families at- risk.

137

Chapter 5 Study 4

5.6 References 1. Stoltenborgh, M., et al., Cultural–geographical differences in the occurrence of child physical abuse? A meta‐analysis of global prevalence. International Journal of Psychology, 2013. 48(2): p. 81-94. 2. Stoltenborgh, M., et al., The prevalence of child maltreatment across the globe: Review of a series of meta‐analyses. Child Abuse Review, 2015. 24(1): p. 37-50. 3. Nadan, Y., J.C. Spilsbury, and J.E. Korbin, Culture and context in understanding child maltreatment: Contributions of intersectionality and neighborhood-based research. Child abuse & neglect, 2015. 41: p. 40-48. 4. Drake, B., S.M. Lee, and M. Jonson-Reid, Race and child maltreatment reporting: Are Blacks overrepresented? Children and youth services review, 2009. 31(3): p. 309-316. 5. Kim, H., D. Chenot, and J. Ji, Racial/ethnic disparity in child welfare systems: A longitudinal study utilizing the Disparity Index (DI). Children and Youth Services Review, 2011. 33(7): p. 1234-1244. 6. Lee, B., et al., Delineating disproportionality and disparity of Asian-Canadian versus White-Canadian families in the child welfare system. Children and Youth Services Review, 2016. 70: p. 383-393. 7. Clarke, J., The challenges of child welfare involvement for Afro-Caribbean families in Toronto. Children and youth services review, 2011. 33(2): p. 274-283. 8. Korbin, J.E., Culture and child maltreatment: Cultural competence and beyond. Child Abuse & Neglect, 2002. 26(6-7): p. 637-644. 9. Elliott, K. and A. Urquiza, Ethnicity, culture, and child maltreatment. Journal of Social Issues, 2006. 62(4): p. 787-809. 10. Alink, L.R., et al., Is elevated risk of child maltreatment in immigrant families associated with socioeconomic status? Evidence from three sources. International Journal of Psychology, 2013. 48(2): p. 117-127. 11. Cardoso, J.B., et al., Nativity and immigration status among Latino families involved in the child welfare system: Characteristics, risk, and maltreatment. Children and Youth Services Review, 2014. 44: p. 189-200. 12. Lee, B., W. Rha, and B. Fallon, Physical abuse among Asian families in the Canadian child welfare system. Journal of Aggression, Maltreatment & Trauma, 2014. 23(5): p. 532-551. 13. Schick, M., et al., Child maltreatment and migration: a population-based study among immigrant and native adolescents in Switzerland. Child maltreatment, 2016. 21(1): p. 3-15. 14. Euser, E.M., et al., Elevated child maltreatment rates in immigrant families and the role of socioeconomic differences. Child Maltreatment, 2011. 16(1): p. 63-73.

138

Chapter 5 Study 4

15. Lhila, A. and K.I. Simon, Prenatal health investment decisions: does the child's sex matter? Demography, 2008. 45(4): p. 885-905. 16. Puri, S., et al., “There is such a thing as too many daughters, but not too many sons”: A qualitative study of son preference and fetal sex selection among Indian immigrants in the United States. Social science & medicine, 2011. 72(7): p. 1169-1176. 17. Ebenstein, A., H. Li, and L. Meng, The impact of ultrasound technology on the status of women in China, in Hebrew University and Tsinghua University Working Paper. 2010. 18. Bharadwaj, P., G.B. Dahl, and K. Sheth, Gender discrimination in the family. The economics of the family: How the household affects markets and economic growth, 2014. 2: p. 237. 19. Brar, A., et al., Sex Ratios at Birth Among Indian Immigrant Subgroups According to Time Spent in Canada. Journal of Obstetrics and Gynaecology Canada, 2017. 39(6): p. 459-464. e2. 20. Ray, J.G., D.A. Henry, and M.L. Urquia, Sex ratios among Canadian liveborn infants of mothers from different countries. Canadian Medical Association Journal, 2012: p. cmaj. 120165. 21. Urquia, M.L., et al., Sex ratios at birth after induced abortion. Canadian Medical Association Journal, 2016. 188(9): p. E181-E190. 22. Urquia, M.L., et al., Variations in male-female infant ratios among births to Canadian- and Indian-born mothers, 1990-2011: a population-based register study. CMAJ open, 2016. 4(2): p. E116. 23. Perron, L., et al., Female genital cutting. Journal of Obstetrics and Gynaecology Canada, 2013. 35(11): p. 1028-1045. 24. Ayyub, R., Domestic violence in the South Asian Muslim immigrant population in the United States. Journal of Social Distress and the Homeless, 2000. 9(3): p. 237-248. 25. Lee, E., Domestic violence and risk factors among Korean immigrant women in the United States. Journal of Family Violence, 2007. 22(3): p. 141-149. 26. Erez, E., Immigration, culture conflict and domestic violence/woman battering. Crime prevention and community safety, 2000. 2(1): p. 27-36. 27. Anitha, S., Legislating gender inequalities: The nature and patterns of domestic violence experienced by South Asian women with insecure immigration status in the United Kingdom. Violence against women, 2011. 17(10): p. 1260-1285. 28. Straus, M.A., Gender symmetry and mutuality in perpetration of clinical-level partner violence: Empirical evidence and implications for prevention and treatment. Aggression and Violent Behavior, 2011. 16(4): p. 279-288. 29. Radford, L., et al., Child abuse and neglect in the UK today. 2011. 30. Stoltenborgh, M., et al., A global perspective on child sexual abuse: meta-analysis of prevalence around the world. Child maltreatment, 2011. 16(2): p. 79-101.

139

Chapter 5 Study 4

31. Meinck, F., et al., Risk and protective factors for physical and sexual abuse of children and adolescents in Africa: a review and implications for practice. Trauma, Violence, & Abuse, 2015. 16(1): p. 81-107. 32. Pulver, A., et al., A scoping review of female disadvantage in health care use among very young children of immigrant families. Social Science & Medicine, 2016. 152: p. 50-60. 33. Wanigaratne, S., et al., The influence of refugee status and secondary migration on preterm birth. J Epidemiol Community Health, 2016. 70(6): p. 622-628. 34. Chiu, M., et al., Describing the linkages of the immigration, refugees and citizenship Canada permanent resident data and vital statistics death registry to Ontario’s administrative health database. BMC medical informatics and decision making, 2016. 16(1): p. 135. 35. Gilbert, R., et al., Child maltreatment: variation in trends and policies in six developed countries. The Lancet, 2012. 379(9817): p. 758-772. 36. Afifi, T.O., Child abuse and adolescent parenting: Developing a theoretical model from an ecological perspective. Journal of aggression, maltreatment & trauma, 2007. 14(3): p. 89-105. 37. Ahrens, K.A., et al., Plurality of Birth and Infant Mortality Due to External Causes in the United States, 2000–2010. American journal of epidemiology, 2017. 185(5): p. 335-344. 38. Luke, B. and M.B. Brown, Maternal risk factors for potential maltreatment deaths among healthy singleton and twin infants. Twin Research and Human Genetics, 2007. 10(5): p. 778-785. 39. Sobsey, D., W. Randall, and R.K. Parrila, Gender differences in abused children with and without disabilities. Child Abuse & Neglect, 1997. 21(8): p. 707-720. 40. Svensson, B., C.G. Bornehag, and S. Janson, Chronic conditions in children increase the risk for physical abuse–but vary with socio‐economic circumstances. Acta Paediatrica, 2011. 100(3): p. 407-412. 41. Kralj, B., Measuring “rurality” for purposes of health-care planning: an empirical measure for Ontario. Ont Med Rev, 2000. 67(9): p. 33-52. 42. Zou, G., A modified poisson regression approach to prospective studies with binary data. American journal of epidemiology, 2004. 159(7): p. 702-706. 43. Zou, G. and A. Donner, Extension of the modified Poisson regression model to prospective studies with correlated binary data. Statistical methods in medical research, 2013. 22(6): p. 661-670. 44. Allison, P.D., Fixed effects regression methods for longitudinal data using SAS. 2005: Sas Institute. 45. D'onofrio, B.M., et al., Familial confounding of the association between maternal smoking during pregnancy and offspring criminality: a population-based study in Sweden. Archives of general psychiatry, 2010. 67(5): p. 529-538.

140

Chapter 5 Study 4

46. Lahey, B.B. and B.M. D’Onofrio, All in the family: Comparing siblings to test causal hypotheses regarding environmental influences on behavior. Current directions in psychological science, 2010. 19(5): p. 319-323. 47. King, B., et al., Factors associated with racial differences in child welfare investigative decision-making in Ontario, Canada. Child abuse & neglect, 2017. 73: p. 89-105. 48. Kaushal, N. and Y. Lu, Recent immigration to Canada and the United States: A mixed tale of relative selection. International Migration Review, 2015. 49(2): p. 479-522. 49. Trocmé, N.M., et al., Major findings from the Canadian incidence study of reported child abuse and neglect. Child abuse & neglect, 2003. 27(12): p. 1427-1439. 50. Gilbert, R., et al., Burden and consequences of child maltreatment in high-income countries. The lancet, 2009. 373(9657): p. 68-81. 51. Carlson, B.E., Children exposed to intimate partner violence: Research findings and implications for intervention. Trauma, Violence, & Abuse, 2000. 1(4): p. 321-342. 52. MacMillan, H.L., E. Jamieson, and C.A. Walsh, Reported contact with child protection services among those reporting child physical and sexual abuse: results from a community survey. Child Abuse & Neglect, 2003. 27(12): p. 1397-1408. 53. Fallon, B., et al., Ontario incidence study of reported child abuse and neglect 2013. 2015: Canadian Child Welfare Research Portal. 54. Schnitzer PG, Slusher PL, Kruse RL, Tarleton MM. Identification of ICD codes suggestive of child maltreatment. Child abuse & neglect, 2011. 35(1): p. 3-17. 55. Scott D, Tonmyr L, Fraser J, Walker S, McKenzie K. The utility and challenges of using ICD codes in child maltreatment research: A review of existing literature. Child abuse & neglect, 2009. 33(11):791-808.

141

Chapter 5 Study 4 Table 5-1 Descriptive characteristics of all eligible children to born non-immigrant and immigrant mothers in Ontario, Canada, 2002-2011 (n=1,240,946). All data are presented as a number (%) unless otherwise specified

Non-immigrant Immigrant mother-child pairs mother-child pairs Standardized Characteristic (N = 925,656) (N = 315,290) difference* Maternal age, years ≤19 9257 (1) 12611 (4) -0.19 20-34 703498 (76) 239621 (76) 0.00 ≥35 212901 (23) 63057 (20) 0.08 Child gender Male 453572 (49) 154494 (49) 0.00 Female 472084 (51) 160796 (51) 0.00 Birth order First 453571 (49) 182870 (58) -0.18 Second 333236 (36) 97739 (31) 0.10 Third 101822 (11) 28376 (9) 0.10 ≥Fourth 37026 (4) 9459 (3) 0.06 Plurality Singleton 892330 (96) 306777 (97) -0.05 Multiple 33326 (4) 8640 (3) 0.05 Neighborhood income quintile 1 (lowest) 166618 (18) 107198 (34) -0.36 2 175875 (19) 72517 (23) -0.10 3 194388 (21) 59906 (19) 0.04 4 212900 (23) 47293 (15) 0.18 5 (highest) 175875 (19) 28376 (9) 0.30 Residential area Urban 805321 (87) 312137 (99) 0.49 Rural 120335 (13) 3153 (1) 0.49 Complex chronic condition Present 46282 (5) 18917 (6) -0.01 Absent 879374 (95) 299526 (95) 0.00 Maternal birthplace India - 43427 (14) N/A China - 29389 (9) Pakistan - 24259 (8) Philippines - 21619 (7) Other South Asia - 35406 (11) Other Western - 25685 (8) North Africa/ Middle East - 24694 (8) Hispanic America - 24610 (8) East Europe - 23814 (8) Sub-Saharan Africa - 22726 (7) Other East Asia/Pacific - 20855 (7) Caribbean - 18806 (6) Time since maternal immigration, years ≤4 - 167105 (53) N/A 5-9 - 72516 (23) 9-14 - 44140 (14) ≥15 - 31529 (10) * Standardized difference = difference in proportions between immigrants and non-immigrants divided by the standard error. A value higher than 0.10 or below -0.10 is accepted as an important difference.

142

Chapter 5 Study 4 Table 5-2 Five-year rates and rate ratios for health-care identified early childhood maltreatment according to maternal birthplace among children born in Ontario Canada between 2002-2011

5-year RR (95%CI) aRR* (95%CI) Rate Per 1000 Immigrant status Non-immigrant (N=925,656) 16 1.00 (ref.) 1.00 (ref.) Immigrant (N=315,290) 10 0.59 (0.57-0.61) 0.64 (0.61-0.66) Maternal birthplace Canada (N=925,656; referent) 16 1.00 (ref.) 1.00 (ref.) India (N=43,427) 8 0.49 (0.44-0.54) 0.53 (0.47-0.59) China (N=29,389) 8 0.49 (0.43-0.56) 0.56 (0.49-0.64) Other East Asia/Pacific (N=20,855) 8 0.48 (0.41-0.56) 0.56 (0.48-0.66) Pakistan (N=24,259) 9 0.53 (0.46-0.60) 0.55 (0.48-0.63) North Africa/ Middle East (N=24,694) 8 0.50 (0.43-0.57) 0.52 (0.45-0.60) East Europe (N=23,814) 10 0.59 (0.52-0.68) 0.71 (0.62-0.81) Other Western (N=25,685) 10 0.61 (0.54-0.69) 0.69 (0.61-0.79) Other South Asia (N=35,406) 10 0.62 (0.56-0.69) 0.66 (0.59-0.73) Hispanic America (N=24,610) 11 0.65 (0.57-0.73) 0.67 (0.60-0.76) Philippines (N=21,619) 9 0.54 (0.47-0.62) 0.60 (0.52-0.70) Sub-Saharan Africa (N=22,726) 12 0.74 (0.66-0.84) 0.73 (0.65-0.83) Caribbean (N=18,806) 16 0.98 (0.88-1.10) 0.94 (0.82-1.06) RR= Rate ratio, aRR=adjusted rate ratio 95% CI= 95% Confidence Interval *Adjusted for maternal age, infant gender, plurality, neighborhood income quintile, rurality, complex chronic conditions, birth order

143

Chapter 5 Study 4 Table 5-3 Five-year rates and rate ratios for sub-components of health-care identified early childhood maltreatment identified comparing children of immigrants to children of non-immigrants born 2002-2011

N maltreated (rate per 1000 children

with a maltreatment-related health visit) RR* (95%CI) Children of non- Children of immigrants immigrants (N = 14,433) (N = 2962) Maltreatment sub-component 1. Maltreatment syndrome 855 (37) 76 (14) 0.37 (0.29-0.47) 2. Assault 479 (21) 105 (19) 0.91 (0.74-1.12) 3. Undetermined cause of injury 6121 (268) 1250 (228) 0.85 (0.81-0.90) 4. Adverse social circumstances 6978 (305) 1531 (279) 0.91 (0.87-0.96) *RR= unadjusted Rate ratio, Referent group is children of non-immigrants 95% CI= 95% Confidence Interval These subcomponents are not mutually exclusive, a small proportion of children may have had more than one type of maltreatment documented over their follow-up period.

144

Chapter 5 Study 4

Table 5-4 Within-family female:male odds ratios for five-year rates of health care identified early childhood maltreatment, by maternal birthplace, among all eligible children with an opposite-gender sibling born in Ontario, Canada between 2002-2011

Female Male F:M F:M N maltreated N maltreated OR aOR (5-year rate per 1000) (5-year rate per 1000) Overall 3082 (14) 3067 (14) 1.00 (0.95-1.06) 1.01 (0.95-1.07) Maternal Birthplace Canada (N = 324,583) * 1.02 (0.96-1.08) 2649 (16) 2622 (16) 1.02 (0.96-1.08) these are good Immigrant Birthplace† India (N=13,134) 48 (7) 53 (8) 0.47 (0.16-1.39) 0.79 (0.50-1.23) China (N=7212) 25 (7) 23 (6) 0.55 (0.08-3.98) 0.81 (0.41-1.59) Philippines (N=5697) 20 (7) 27 (9) 0.63 (0.29-1.37) 0.63 (0.29-1.37) Other Western (N=8297) 35 (8) 41 (10) 0.69 (0.41-1.16) 0.69 (0.41-1.16) Pakistan (N=9363) 41 (9) 33 (7) 0.76 (0.44-1.30) 0.76 (0.44-1.30) Hispanic America (N=6985) 29 (8) 28 (8) 0.85 (0.42-1.71) 0.85 (0.42-1.71) Caribbean (N=5375) 40 (15) 41 (15) 0.96 (0.56-1.63) 0.96 (0.56-1.64) Other South Asia (N=12,092) 56 (9) 50 (4) 1.01 (0.65-1.56) 1.01 (0.65-1.56) North Africa/ Middle East 44 (9) 42 (9) 1.10 (0.65-1.84) 1.10 (0.65-1.84) (N=9278) Sub-Saharan Africa (N=8766) 48 (11) 48 (11) 1.14 (0.71-1.82) 1.14 (0.71-1.82) Other East Asia/Pacific 25 (9) 23 (8) 1.44 (0.62-3.37) 1.44 (0.62-3.67) (N=5761) East Europe (N=6337) 32 (10) 26 (8) 1.55 (0.80-2.99) 1.55 (0.80-2.99) OR=odds ratio, aOR= adjusted odds ratio, 95% CI= 95% Confidence Interval * aOR adjusted for income quintile, maternal age, rurality, complex chronic conditions, birth order † aOR adjusted for income quintile, maternal age, rurality, complex chronic conditions, birth order, duration of residence in Canada

145

Chapter 5 Study 4

Figure 5-1 Flow Chart of the Primary Study Cohort and Sibling Sub-Cohort

Live Births in Ontario, Canada Children excluded due to loss of provincial April 1st 2002 to March 31st 2011 health coverage from moving or death N=1,315,186 N=28,194

Children eligible for provincial health Children excluded due to loss of contact insurance for 60 months after birth N=38,500 N=1,286,992

Children with 60 months of complete Children excluded due to missing follow-up after birth until March 31st 2016 covariate data N=1,248,492 N=6145

Final Primary Study Cohort N=1,240,946

Sub-cohort of eligible children with an opposite gender sibling N=422,880 children of 184,982 mothers

146

Chapter 5 Study 4

5.7 Appendices Appendix 5-1. The method used to identify child maltreatment

VARIABLE DESCRIPTION Binary variable (Yes/No) indicating at least one occurrence of At least one fatal or non-fatal child maltreatment identified in the emergency diagnostic code room, in hospital, or out of hospital deaths occurring between 0 Y/N in 1, 2, 3, 4, wks and up to 60 months of age 5 Description Data Source

1. Either fatal or non-fatal maltreatment syndrome: T74, NACRS, CIHI- Y06, Y07, T73 DAD

2. OR fatal or non-fatal assault: X85- Y09 NACRS, CIHI- DAD 3. OR fatal or non-fatal undetermined cause: Y10-Y34, NACRS, CIHI- Y87, Z04.0, Z04.5, Z04.8 DAD 4. OR fatal or non-fatal adverse social circumstances: NACRS, CIHI- Z60-Z63, Z72, Z74, Z76.1, Z76.2, Z81, Z86.5, Z91.6, DAD Z91.8 5. OR Deaths out of hospital associated with Ontario Registrar maltreatment, where the cause of death is coded as any General of the above: Database- Vital i. Maltreatment syndrome: T74, Y06, Y07, T73 Statistics ii. Assault (homicides and injury purposively inflicted): X85- Y09 iii. Undetermined cause (other violence): Y10- Y34, Y87, Z04.0, Z04.5, Z04.8 iv. Adverse social circumstances: Z60-Z63, Z72, Z74, Z76.1, Z76.2, Z81, Z86.5, Z91.6, Z91.8

147

Chapter 6 Discussion

This concluding chapter is divided into six subsections. First, I return to the overarching aim of the dissertation and summarize the main findings of the research studies to address this aim.

Second, I situate my findings in the broader literature of gender inequality and the health and well-being of immigrant children and highlight the research contributions. Third, I extrapolate from my research contributions to offer some suggestions for research. Fourth, I make some recommendations for practice, with the goal of promoting early child health care services and reducing the occurrence of early childhood maltreatment, for all children, as well as for vulnerable groups. Fifth, I describe some of the more significant limitations of my research studies in greater depth than the space afforded in the previous chapters. Finally, I describe the key learnings as a doctoral candidate in epidemiology, and briefly summarize the conclusions of my work.

6.1 Summary of findings

I aimed to improve our understanding of the health of children in immigrant families by considering how gender inequality around the world may influence early child health and well- being within a high immigration context. The four studies in my dissertation correspond to three distinct but related research objectives addressing this aim. The principal substantive finding is that maternal birthplace has significant associations with child health and well-being indicators, including both routine preventive health services and early childhood maltreatment, but that gender inequality has minimal influence on early childhood health and wellbeing, except for the select case of children whose mothers were born in India. While numerous studies in many source countries show gender disparities affecting children’s health, gender inequity in

148

Chapter 6 Discussion immigrant source countries do not appear to affect routine primary care and maltreatment in early life for most groups. Overall, my research findings are encouraging as they demonstrate that the children of immigrants are well cared for within the Canadian context.

In Study 1 (Chapter 2) I mapped the literature regarding gender inequities in health care use among children of parents who migrate to Western, high-income countries. The main finding was that health care use among children in immigrant families might differ between boys and girls, but the reasons are unexplored and potentially due to differences in health care need. I also found few studies designed with a gender-based analysis in mind.

Next, in Study 2 (Chapter 3) I compared the prevalence of two markers of ‘health-care neglect’—under-immunization and inadequate well-child visits across maternal birthplaces in comparison to mothers born in Canada. To further determine if gender disparities in these outcomes exist, in Study 3 (Chapter 4), I compared the odds of health-care neglect outcomes between daughters and sons in each maternal birthplace. Children born to most immigrant women are at reduced risk for inadequate routine health care and early childhood maltreatment than children of non-immigrants, with some exceptions. Children of all immigrant mothers except those of Somali-, Polish-, and American-born mothers are at reduced risk of under- immunization compared to those of Canadian-born mothers. While children of immigrant mothers overall had more immunizations than children of non-immigrant mothers, I observed less frequent well-child visits in 50% of the studied MCOB, including India among six other countries. Somalia emerged as a maternal birthplace where children were at markedly elevated risk of sub-optimal well-child care compared to Canada (9% and 84% elevated risk for under- immunization and inadequate well-child visits, respectively).

149

Chapter 6 Discussion

In general, all maternal birthplaces achieved within-family gender equity in routine health care.

However, I observed that daughters of Indian-born mothers and specifically those whose mother tongue is Punjabi had higher odds of inadequate well-child visits compared to their brothers.

Taken together, girls with mothers born in India appear to encounter a uniquely intersecting disadvantage for well-child visits, associated with their maternal birthplace and compounded with their gender. A similar effect was observed in unadjusted models for Afghanistan, however it did not reach statistical significance following adjustment. A significant effect may have been more difficult to detect for Afghanistan than India, due to the relatively smaller population.

In my final dissertation study (Study 4, Chapter 5), I looked to more severe forms of early childhood adversity to identify differences in early childhood maltreatment by maternal birthplace and disparities by child gender. I demonstrated that children of immigrants are significantly less likely to experience maltreatment in early childhood than children of non- immigrants. Children of Caribbean-born mothers did not differ from the non-immigrant group.

No significant differences were seen between daughters and sons in the odds of early childhood maltreatment by maternal birthplace.

6.2 Research Contributions

The inconclusive findings of a recent systematic review demonstrated the insufficiency of the immigrant versus non-immigrant distinction in characterizing differences in immigrant children’s health care use [1]. In that review, most included studies showed lower vaccination rates and lower primary care use among immigrant children. Even research studies situated in

Canada report significant difficulties for immigrants accessing and underutilization of health care services [2-4]. My research demonstrates that immigrant and non-immigrant differences in child

150

Chapter 6 Discussion health and well-being likely depend significantly on the maternal country of origin. While I did find a lower number of well-child visits compared to Canada for seven of the maternal birthplace groups, five countries did not differ, and two countries had a high number. Moreover, despite a lower number of visits compared to Canada, there was a distinct pattern of a higher number of immunizations in eleven maternal birthplace groups.

There has been some qualitative research aiming to understand both immigrant and non- immigrant parent’s perceptions and barriers to child well-being and pediatric health care services. Condon and colleagues conducted focus groups with Romanian, Polish, Somali and

Pakistani parents to better understand barriers to child well-being in the immigrant context in the

United Kingdom. Parents felt that barriers to child well-being arose due to difficulty finding work and childcare, experiencing discrimination, less time for breastfeeding, poor access to nutritional foods, loss of support and community traditions, and fewer opportunities for active play [5]. A second qualitative study focused on knowledge and beliefs about preventive health care among Somali women in the United States. Participants stated that many Somali women tend to stay at home and delay seeking medical care until they are very ill, either due to: 1) not knowing what would happen during the medical encounter; or 2) a lack of knowledge about the appropriate use of health care services. While participants were familiar with the concept of periodic health examination, most said they would not go to the doctor if they were not sick [6].

The findings of Study 4 similarly provide a more nuanced understanding of child maltreatment by parental nativity rather than solely by ethnicity, race, or immigration. To my knowledge, until now, limited research has delineated racial or ethnic differences in child maltreatment by parental nativity. Adjustment for sociodemographic characteristics did not account entirely for reduced risk of maltreatment among immigrant groups. This suggests that differences in child

151

Chapter 6 Discussion services involvement across ethno-racial groups may not be driven solely by socioeconomic factors, after considering immigration. Research using individual socioeconomic position rather than neighborhood income quintile may however have elicited different results.

My dissertation studies point to maternal birthplace as a social determinant of child health outcomes, but Study 3 also implies that gender inequity in one’s birthplace can have associated consequences for children in immigrant families. Although to my knowledge, no other population-based longitudinal studies have been conducted to look at gender disparities in child health, our results are supported by numerous studies that have identified gender bias in family planning in Indo-Canadian families, evidenced by skewed sex ratios at birth due to sex-selective termination [7-10]. I found gender disparities (son-preference) in well-child visits at 24 months, which demonstrates that the effects of son preference may extend beyond the family planning stage. The large body of literature identifying significant female disadvantage in healthcare globally supports the findings of Study 3. Taken together, researchers, policy-makers, and stakeholders alike should not assume that gender inequalities in these outcomes do not exist in high-income, seemingly gender-equal societies, and that it is a matter worthy of empirical investigation.

6.3 Implications for Future Research There are several research implications stemming from my doctoral research studies. First, the well-being of immigrant children merits further specific research attention to the role of parental birthplace. While existing studies of child health take into account and highlight the importance of either immigration or race or ethnicity, researchers should exhibit greater sensitivity to the role of parental birthplace specifically. The primary benefit of disaggregating immigrant groups

152

Chapter 6 Discussion by birthplace is the ability to uncover several associations that would otherwise be masked by the binary categories of immigrant versus non-immigrant.

Qualitative research within specific immigrant communities that replicates earlier findings would be helpful to understand if similar beliefs exist for health care seeking for children in the

Canadian context. Moreover, further research to identify whether reduced visits have clinical implications for the development and well-being of children of immigrants is warranted.

A logical next step would be to stratify analyses by immigrant and non-immigrant groups (and by maternal birthplace, sample size allowing), to test whether maternal immigration modifies associations with sociodemographic factors. For example, socioeconomic position is associated with the occurrence for early childhood maltreatment and abuse [11-13]. However socioeconomic gradients in early childhood maltreatment may differ for children of non- immigrants compared to children of immigrants (or for children of Caribbean-born women in particular). I am currently pursuing a further analysis to examine how socioeconomic and family size gradients affect the outcomes I have studied. Research findings can offer clues as to the mechanisms behind observed maternal birthplace differences in indicators of maltreatment neglect.

Moreover, analyses focused on immigrants specifically allow for closer study of immigrant- specific factors, such as immigration class (refugee, economic, family migrant), as well as time since migration.

As I discussed in the introduction chapter of this thesis (Chapter 1), scholars have proposed a life-cycle of gender discrimination affecting women in South Asia (Chapter 1, Figure 1-2) [14].

This thesis work helps further build a knowledge base that highlights vulnerable time points for

153

Chapter 6 Discussion gender discrimination in the immigration context, with evidence for sex-selective abortion and

‘benign’ neglect in infancy/poor health care seeking. My dissertation studies show that research is warranted that takes a gender-based lens to study other forms of child well-being, including at other points in the lifecycle. For example, studies are warranted on educational outcomes, mental health, and the experience of violence in later childhood and adolescence.

My approach to gender-based analysis contributes a guiding methodology from which to study and monitor gender inequality. Gender-based analysis typically disaggregates outcomes by gender, perhaps stratifying risk by gender, or comparing one gender to another [15, 16]. The fixed-effects analytic approach I took in studies 3 and 4 offers a new application for the methodological and statistical approach whereby siblings are compared while controlling for common causes, diminishing the influence of unmeasured confounding. It is particularly useful for studies that aim to compare outcomes between male and female siblings, where family factors are shared [17, 18]. This approach more directly estimates differences due to gender disparities, meaning unfair, avoidable differences [19]. In the absence of other causes such as underlying illness, family composition, or health behaviors, and environment, there should be no difference between males and female children. While I found no sibling gender differences for early childhood maltreatment in any maternal birthplace, future studies can use a similar approach to examine gender disparities in maltreatment as well as replicate analyses for health care service use among school-age children. Moreover, there may be other settings where a similar approach can be valuable to evaluate gender inequity, including in a workplace or school setting where the aim is to control for shared environmental factors and estimate the main effect of gender. I hope that more knowledge surrounding gender inequity is generated using this approach and help to yield a more robust evidence base in support of child health and well-being.

154

Chapter 6 Discussion

6.4 Clinical and Public Health Implications

6.4.1 Ensuring adequate preventive routine care

Although my dissertation studies were encouraging in that children of most immigrants were well-protected from under-immunization and healthcare identified maltreatment, my findings also highlight several practical implications to ensure early childhood health and well-being both within Ontario as well as other high-income, high-immigration settings. First, practitioners may work to encourage all parents to schedule and attend the five minimum required well-child visits by the age of two (the Canadian Pediatric Society currently recommends six, but the 15-month visit is currently considered optional) [20]. Because I observed inadequate visits compared to

Canada in only 50% of maternal birthplaces, recognizing maternal nativity may help healthcare practitioners identify families who may have unmet health care needs. Policies and programs that address some of the commonly reported barriers [4, 21-23] to health care for immigrants may help to ensure adequate well-child visit scheduling. Betancourt and colleagues offer a cultural competence framework for the implementation of measures to address racial/ethnic disparities in health and healthcare, including low preventive health care utilization [24]. The cultural competence framework sees both “social factors” (e.g., socioeconomic status, supports/stressors, environmental hazards) and “cultural factors” as having essential influences on patients' experience of health care. Barriers are categorized into sociocultural barriers to care at the organizational (leadership/workforce), structural (processes of care), and clinical (provider- patient encounter) levels. Interventions to break through barriers that may impede adequate well- child visits include minority recruitment into the health professions, development of interpreter services and language-appropriate health educational materials, and provider education on cross-

155

Chapter 6 Discussion cultural issues [24]. Additional instrumental supports to improve access to affordable transportation and child care may also reduce some of the differences I observed in the number of well-child visits across maternal birthplace groups. Children of Somalia-born mothers emerged as the group at most significant risk for inadequate well-child visits and under- immunization. Qualitative research should be conducted in the Canadian setting to assess health beliefs among Somali women regarding the use of preventive health care services for children. If further research replicated the European and American findings described above [6, 25], building partnerships with Somali community health organizations could help to develop targeted messaging to change beliefs around child health care-seeking behavior [26].

Second, both clinical and public health practitioners should continue to draw attention to the importance of early vaccination. Canada currently has a goal of 95% vaccination coverage at 24 months by 2025 [27, 28]. While dose-counting using billing data likely underestimates true immunization coverage [29], our findings coupled with others using timing and antigen-specific coverage approaches [27-31] show that Canada’s immunization goal is far from being achieved.

Additionally, it is clear from our studies that there are large differences in dose administration across varied maternal birthplace groups. Complete vaccination of children of immigrants is of particular importance because they may be at increased risk of vaccine-preventable disease due to frequent travel to parental home countries [32]. Ensuring adequate vaccinations is vital for the general population, children of non-immigrants, as well as other groups at elevated risk of low vaccinations, including children of Poland, USA, and Somalia-born parents. There are many existing initiatives underway to counter vaccine hesitancy (increasing knowledge and changing attitudes). Systematic review evidence shows that multicomponent and dialogue-based interventions targeting specific under-vaccinated populations are most effective, especially those

156

Chapter 6 Discussion tailored to specific populations and their specific concerns [33]. The review also highlighted the importance of mandating vaccinations, improving knowledge, awareness, access, and convenience, as well as engaging with religious and other individuals in leadership positions.

One example of a tailored community-based intervention is ongoing within the Somali diaspora in Sweden and the United States and involves targeted Somali-language engagement and video messaging in healthcare settings [26]. Encouraging vaccines in early childhood can also prevent children from being denied access to school in jurisdictions where vaccines are mandated (e.g.,

Ontario’s Immunization of School Pupils Act) [34].

6.4.2 Ensuring gender equity in child routine healthcare

The findings of Study 3 and 4 are encouraging in that I observed gender equity in health-care- identified maltreatment and routine childhood primary care for almost all groups. However, my findings still support the need for continued discussions and programs aimed at reducing gender inequality in the Indo-Canadian community. Efforts may also be warranted among the general immigrant population, who migrated from countries with high gender inequality at the time of migration, as measured by the UN Gender Inequality Index [35]. Some members of my PhD dissertation committee and I were recently involved in a scoping review to map the literature concerning interventions to reduce adverse health outcomes resulting from manifestations of gender bias among immigrant populations [36]. While we found 29 reports of interventions addressing gender bias, almost all were based within Latino communities in the United States and were focused on reducing domestic violence [36]. Although the studies were limited in scope, it is encouraging that many of them included evaluations and reported achieving their objective. The results of that review imply that interventions to improve gender equity are

157

Chapter 6 Discussion worthwhile and can be useful. There is ongoing work in Ontario by the South Asian Women’s

Centre and Punjabi Community Health Services that is raising awareness surrounding sex- selective abortion and supporting researcher-community collaborations to better understand and address son preference [37]. My next step is to continue building partnerships with these organizations among others to draw attention to female disadvantage in early child health visits.

6.5 Methodological Considerations

The previous chapters discussed limitations for each of the four included studies. Here I devote attention to some additional constraints.

6.5.1 The use of administrative healthcare data for outcome measurement

Studies 2 to 4 used administrative healthcare data to ascertain each outcome measurement including the number of immunization doses, the number of well-child visits, and the occurrence of early childhood maltreatment. OHIP data for immunizations and well-child visits provides a high level of specificity [38], but likely underestimates the correct outcome count, as discussed in chapters 3 and 4. Nurse practitioners have an expanding scope of practice and most do not submit billing data. As well, physicians working at community health centers are salaried.

Ultimately, primary care administrative data relies on accurate physician billing. There has been substantial variation in physician billing for primary care over the time periods in studies 2 and 3

(Chapters 3 and 4). 2008 saw the introduction of new primary care funding models that moved away from fee-for-service payments to a patient rostering model. Some physicians may also be paid by a blended model of fee-for-service and capitation. Physicians under these new funding models are only paid an incentive fee for achieving certain levels of immunization coverage in their practice. By excluding children with no billings for immunizations or well-child visits, as

158

Chapter 6 Discussion well as those seen at Community Health Centres, one can be reasonably confident that included children were indeed cared for by billing health care practitioners. However, this resulted in a non-negligible excluded population and may have introduced some selection bias to studies 2 and 3. I observed that children excluded were most likely to live in the lowest neighborhood income quintile, have adolescent mothers, live in a rural area, and have a large family size.

Moreover, children in Somali families had the highest proportion of excluded children due to

CHC care and no documented visits. If health care utilization data were available for all children

(including those with the most adverse sociodemographic profile), I might have observed even more pronounced effect estimates, or more attenuated estimates if the excluded children were indeed well-immunized.

Another limitation of administrative data common to many research studies is that outcome identification necessitates health system contact to provide the opportunity for diagnosis. While immunizations and well-child visits are a routine experience for almost every child, the occurrence of early childhood maltreatment and abuse remains relatively uncommon. In study 4,

I was only able to capture maltreatment identified within a healthcare setting. Hospital and emergency department visits may only capture a small proportion of true maltreatment events

[39], especially if the maltreatment does not result in injury necessitating swift medical attention.

Child maltreatment is also commonly identified through mental health services, schools, social services, and law enforcement [40]. Some injuries identified in healthcare settings may only be deemed non-accidental after involvement from police or child protective services. In very early childhood when there is limited interaction between children and others outside the family setting, maltreatment may not be identified at all. Patterns of maltreatment identification could vary by immigration as well as by country of origin, potentially contributing to differential

159

Chapter 6 Discussion outcome misclassification, resulting in biased effect estimates either towards or away from the null effect.

6.5.2 Unmeasured and Residual Confounding

While studies 2 through 4 each accounted for several important confounding factors, there remain several sources of potential unmeasured and residual confounding due to both data limitations as well as methodological choices. First, I chose not to include variables on maternal education, English language ability, and marital status available from the IRCC recorded at landing to Canada. These factors may have changed substantially depending on her age at arrival and time until delivery for an index child. Second, I was unable to identify additional children who may have been born to a given mother before arrival in Canada. Those children could have comprised sibling sets included in studies 3 and 4 (chapters 4 and 5), but unfortunately could not be identified as potential siblings. However, maternal parity recorded at the delivery episode was used to derive the included child’s birth order. Therefore, all analyses accounted for accurate family size.

Due to selecting Canada as the comparator, I could not study relevant immigrant-specific variables including immigrant class (refugee, economic, family reunification, among other classes), duration of residence in Canada since landing, and English language ability. Selecting an internal immigrant reference group (i.e., immigrants from high-income countries) is therefore a crucial next step because: 1. routine care and child maltreatment may vary significantly depending on the duration of residence in Canada and refugee status and 2. the distribution of these factors varies across birthplace groups. From an etiologic perspective, confounder measurement is of chief importance. However, from a public health and clinical perspective, the

160

Chapter 6 Discussion crude counts, prevalence, and rates reported in my dissertation studies remain useful—we are now able to recognize groups of children being underserved in the health care system and observe differences in maltreatment rates. This limitation, however, does not apply to the gender disparity analyses in studies 3 and 4 (Chapters 4 and 5), as familial factors, including maternal immigration characteristics, are controlled for using a fixed-effects approach [18].

Finally, my dissertation studies did not corroborate paternal and maternal nativities, as I could not link fathers to their children with the current data. Stronger associations with both birthplace differences and gender inequalities may exist among parents with the same nativity compared to parents with different nativities, particularly if one is Canadian-born. Future research can seek additional data sources to reclassify children with this in mind.

6.5.3 Generalizability

The findings from my dissertation studies are generalizable to other high immigration, high- income country contexts with similar diversity and predominantly voluntary migration [41].

However, findings may not be generalizable to settings where immigrants may not have equivalent health care insurance. Canada differs from other high immigrant-receiving countries, like the United States, due to its highly selective immigration process. In Canada, immigration strongly favors well-educated immigrants proficient in the host language [42], which may help to explain the relatively lower maltreatment rates and higher vaccination rates I observed. For example, English language facility and education can support health care access among immigrants [21] and higher education levels may be associated with lower childhood maltreatment. Additionally, the findings of this dissertation may not apply to immigrants who are

161

Chapter 6 Discussion not permanent residents of Canada. That may include very newly arrived immigrants as well as those who are undocumented.

6.6 Critical learnings for the doctoral candidate

These four dissertation studies offered me numerous training opportunities to emerge as an independent scientist in epidemiology. In the first scoping review study, I was responsible for synthesizing the extensive literature on immigrant children’s health to identify the use of gender- based analysis to study health inequalities. The three remaining empirical studies required me to learn how to assess and use population-based administrative data, particularly in measuring pediatric primary care use. Throughout the years working on these projects, I collaborated with my advisory committee to use iterative and creative problem-solving approaches to address significant data limitations, mainly through the ascertainment of primary care data. I developed a thorough understanding of primary care funding models and shifts as well as reasonable techniques for immunization data correction, all within the dynamic immunization measurement landscape in Canada.

As well, the data I used to achieve dissertation objectives 2 and three inherently had a nested quality where multiple children may be born to one mother. I studied and consulted with my committee and statisticians on how to best approach this challenge, both conceptually, as well as in statistical analysis software (Study 2). I also had the chance to apply an interesting statistical method, the fixed-effects approach, to directly compare male and female siblings. The fixed- effects analysis allowed me to directly estimate the causal effect of son preference on gender differences in health and well-being. Lastly, I had the opportunity to link new aggregate data from the UN with ICES data, including the UN Gender Inequality Index, as well as other

162

Chapter 6 Discussion measures of gender inequality that I explored. Overall, my dissertation has allowed me to develop a body of work to which I applied a robust epidemiologic toolkit and helped advance the evidence base for children’s health and well being.

6.7 Conclusions

In this dissertation, I pursued a series of inquiries to identify differences in routine preventive health care and maltreatment driven by maternal birthplace and child gender. I found that parental birthplace is significantly associated with early childhood routine preventive health care and maltreatment and abuse in a high immigration context. Children of most immigrants are protected from under-immunization and health-care identified maltreatment when compared to children of non-immigrants. Canada’s universal health care system combined with immigrant parents’ health care decision-making appears to support the healthcare and wellbeing of children of immigrants. I did find however that some children of immigrants have fewer well-child visits than infants of Canadian-born mothers.

Due to cultural and systemic disadvantages, girls in many world regions receive suboptimal preventative care and are more likely than boys to suffer from forms of benign and severe neglect. Literature exploring the contribution of child gender to immigrant and non-immigrant differences in child health and well-being is extremely limited. My dissertation findings are encouraging from the perspective of gender equity in child healthcare and wellbeing. I demonstrated that gender equity in routine preventive health care is being achieved among children of immigrants. There were no observed gender inequities in maltreatment. However, infant daughters of Punjabi-speaking Indian-born mothers, appear to be at a disadvantage for

163

Chapter 6 Discussion well-child visits compared to their brothers. Effects of son preference and patriarchy may persist postnatally in the Indo-Canadian community.

Taken together, my dissertation makes substantial contributions to both the child health and well- being and gender equity bodies of literature. Canada admits over 250,000 immigrants per year, and that number is expected to increase as immigration becomes the sole contributor to population growth [43]. Therefore, children in immigrant families are a very rapidly growing population. This dissertation generates scientific knowledge and offers methodological advances to support equitable healthcare and wellbeing for all children in Canada and other high immigrant-receiving settings.

164

Chapter 6 Discussion

6.8 References 1. Markkula, N., et al., Use of health services among international migrant children–a systematic review. Globalization and health, 2018. 14(1): p. 52. 2. McKeary, M. and B. Newbold, Barriers to care: The challenges for Canadian refugees and their health care providers. Journal of Refugee Studies, 2010. 23(4): p. 523-545. 3. Newbold, B., Health status and health care of immigrants in Canada: a longitudinal analysis. Journal of health services research & policy, 2005. 10(2): p. 77-83A. 4. Asanin, J. and K. Wilson, “I spent nine years looking for a doctor”: exploring access to health care among immigrants in Mississauga, Ontario, Canada. Social science & medicine, 2008. 66(6): p. 1271-1283. 5. Condon, L. and S. McClean, Maintaining pre-school children's health and wellbeing in the UK: a qualitative study of the views of migrant parents. Journal of Public Health, 2016. 39(3): p. 455-463. 6. Carroll, J., et al., Knowledge and beliefs about health promotion and preventive health care among Somali women in the United States. Health care for women international, 2007. 28(4): p. 360-380. 7. Brar, A., et al., Sex Ratios at Birth Among Indian Immigrant Subgroups According to Time Spent in Canada. Journal of Obstetrics and Gynaecology Canada, 2017. 39(6): p. 459-464. e2. 8. Ray, J.G., D.A. Henry, and M.L. Urquia, Sex ratios among Canadian liveborn infants of mothers from different countries. Canadian Medical Association Journal, 2012: p. cmaj. 120165. 9. Urquia, M.L., et al., Sex ratios at birth after induced abortion. Canadian Medical Association Journal, 2016. 188(9): p. E181-E190. 10. Urquia, M.L., et al., Variations in male-female infant ratios among births to Canadian- and Indian-born mothers, 1990-2011: a population-based register study. CMAJ open, 2016. 4(2): p. E116. 11. Alink, L.R., et al., Is elevated risk of child maltreatment in immigrant families associated with socioeconomic status? Evidence from three sources. International Journal of Psychology, 2013. 48(2): p. 117-127. 12. Euser, E.M., et al., Elevated child maltreatment rates in immigrant families and the role of socioeconomic differences. Child Maltreatment, 2011. 16(1): p. 63-73. 13. Nadan, Y., J.C. Spilsbury, and J.E. Korbin, Culture and context in understanding child maltreatment: Contributions of intersectionality and neighborhood-based research. Child abuse & neglect, 2015. 41: p. 40-48. 14. Fikree, F.F. and O. Pasha, Role of gender in health disparity: the South Asian context. BMJ: British Medical Journal, 2004. 328(7443): p. 823.

165

Chapter 6 Discussion

15. Bryant, T., C. Leaver, and J. Dunn, Unmet healthcare need, gender, and health inequalities in Canada. Health policy, 2009. 91(1): p. 24-32. 16. Greyson, D.L., A.R. Becu, and S.G. Morgan, Sex, drugs and gender roles: mapping the use of sex and gender based analysis in pharmaceutical policy research. International journal for equity in health, 2010. 9(1): p. 26. 17. D'onofrio, B.M., et al., Familial confounding of the association between maternal smoking during pregnancy and offspring criminality: a population-based study in Sweden. Archives of general psychiatry, 2010. 67(5): p. 529-538. 18. Allison, P.D., Fixed effects regression methods for longitudinal data using SAS. 2005: Sas Institute. 19. Braveman, P. and S. Gruskin, Defining equity in health. Journal of Epidemiology & Community Health, 2003. 57(4): p. 254-258. 20. Society, C.P. Schedule of well-child visits 2016 [cited 2018 May 30 2018]; Available from: https://www.caringforkids.cps.ca/handouts/schedule_of_well_child_visits. 21. Flores, G., M. Abreu, and S.C. Tomany-Korman, Limited English proficiency, primary language at home, and disparities in children's health care: how language barriers are measured matters. Public health reports, 2005. 120(4): p. 418-430. 22. Pottie, K., et al., Language proficiency, gender and self-reported health: an analysis of the first two waves of the longitudinal survey of immigrants to Canada. Canadian Journal of Public Health/Revue Canadienne de Sante'e Publique, 2008: p. 505-510. 23. Stella, M.Y. and G.K. Singh, Household language use and health care access, unmet need, and family impact among CSHCN. Pediatrics, 2009. 124(Supplement 4): p. S414- S419. 24. Betancourt, J.R., et al., Defining cultural competence: a practical framework for addressing racial/ethnic disparities in health and health care. Public health reports, 2016. 25. Pavlish, C.L., S. Noor, and J. Brandt, Somali immigrant women and the American health care system: discordant beliefs, divergent expectations, and silent worries. Social Science & Medicine, 2010. 71(2): p. 353-361. 26. Jama, A., et al., Tailored communication interventions targeting Somali community in Sweden regarding MMR vaccinationAsha Jama. European Journal of Public Health, 2017. 27(suppl_3): p. ckx189.047-ckx189.047. 27. Canada, P.H.A.o., Vaccine coverage in Canadian children: Results from the 2013 childhood National Immunization Coverage Survey (cNICS). 2016. 28. Canadian Immunization Registry Network. National Standards for Immunization Coverage Assessment: Recommendations from the Canadian Immunization Registry Network, in Canadian Communicable Disease Report. 2012.

166

Chapter 6 Discussion

29. Wilson, S.E., et al., Methods used for immunization coverage assessment in Canada, a Canadian Immunization Research Network (CIRN) study. Human vaccines & immunotherapeutics, 2017. 13(8): p. 1928-1936. 30. Dummer, T., et al., Immunization completeness of children under two years of age in Nova Scotia, Canada. Can J Public Health, 2012. 103(5): p. e363-7. 31. Guttmann, A., et al., Immunization coverage among young children of urban immigrant mothers: findings from a universal health care system. Ambulatory Pediatrics, 2008. 8(3): p. 205-209. 32. Bacaner, N., et al., Travel medicine considerations for North American immigrants visiting friends and relatives. Jama, 2004. 291(23): p. 2856-2864. 33. Jarrett, C., et al., Strategies for addressing vaccine hesitancy–A systematic review. Vaccine, 2015. 33(34): p. 4180-4190. 34. Mah, C.L., et al., Compulsory school-entry vaccination laws and exemptions: who is opting out in Ontario and why does it matter? Healthcare Policy, 2010. 5(4): p. 37. 35. Gaye, A., et al., Measuring key disparities in human development: The gender inequality index. Human Development Research Paper, 2010. 46: p. 1-37. 36. Januwalla, A., et al., Interventions to reduce adverse health outcomes resulting from manifestations of gender bias amongst immigrant populations: a scoping review. BMC Women's Health, 2018. 18(1): p. 104. 37. Wanigaratne, S.M., Baldev; Kaur, Amandeep; Sekhar, Kripa, Improving health and well- being in Indian immigrant families: a dialogue and pilot educational intervention. 2018. 38. Schwartz, K.L., et al., Validation of infant immunization billing codes in administrative data. Human vaccines & immunotherapeutics, 2015. 11(7): p. 1840-1847. 39. Fallon, B., et al., Ontario incidence study of reported child abuse and neglect 2013. 2015: Canadian Child Welfare Research Portal. 40. Gilbert, R., et al., Recognising and responding to child maltreatment. The Lancet, 2009. 373(9658): p. 167-180. 41. Ferrer, A.M., G. Picot, and W.C. Riddell, New directions in immigration policy: Canada's evolving approach to the selection of economic immigrants. International Migration Review, 2014. 48(3): p. 846-867. 42. Kaushal, N. and Y. Lu, Recent immigration to Canada and the United States: A mixed tale of relative selection. International Migration Review, 2015. 49(2): p. 479-522. 43. Statistics Canada. (2018). "Population growth: Migratory increase overtakes natural increase." from https://www150.statcan.gc.ca/n1/pub/11-630-x/11-630-x2014001- eng.htm.

167